You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
rocksdb/cache/compressed_secondary_cache_...

648 lines
24 KiB

Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
// Copyright (c) 2011-present, Facebook, Inc. All rights reserved.
// This source code is licensed under both the GPLv2 (found in the
// COPYING file in the root directory) and Apache 2.0 License
// (found in the LICENSE.Apache file in the root directory).
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
#include "cache/compressed_secondary_cache.h"
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
#include <algorithm>
#include <cstdint>
#include "memory/jemalloc_nodump_allocator.h"
#include "memory/memory_allocator.h"
#include "rocksdb/convenience.h"
#include "rocksdb/secondary_cache.h"
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
#include "test_util/testharness.h"
#include "test_util/testutil.h"
#include "util/compression.h"
#include "util/random.h"
namespace ROCKSDB_NAMESPACE {
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
class CompressedSecondaryCacheTest : public testing::Test {
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
public:
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
CompressedSecondaryCacheTest() : fail_create_(false) {}
~CompressedSecondaryCacheTest() {}
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
protected:
class TestItem {
public:
TestItem(const char* buf, size_t size) : buf_(new char[size]), size_(size) {
memcpy(buf_.get(), buf, size);
}
~TestItem() {}
char* Buf() { return buf_.get(); }
size_t Size() { return size_; }
private:
std::unique_ptr<char[]> buf_;
size_t size_;
};
static size_t SizeCallback(void* obj) {
return reinterpret_cast<TestItem*>(obj)->Size();
}
static Status SaveToCallback(void* from_obj, size_t from_offset,
size_t length, void* out) {
TestItem* item = reinterpret_cast<TestItem*>(from_obj);
const char* buf = item->Buf();
EXPECT_EQ(length, item->Size());
EXPECT_EQ(from_offset, 0);
memcpy(out, buf, length);
return Status::OK();
}
static void DeletionCallback(const Slice& /*key*/, void* obj) {
delete reinterpret_cast<TestItem*>(obj);
obj = nullptr;
}
static Cache::CacheItemHelper helper_;
static Status SaveToCallbackFail(void* /*obj*/, size_t /*offset*/,
size_t /*size*/, void* /*out*/) {
return Status::NotSupported();
}
static Cache::CacheItemHelper helper_fail_;
Cache::CreateCallback test_item_creator = [&](const void* buf, size_t size,
void** out_obj,
size_t* charge) -> Status {
if (fail_create_) {
return Status::NotSupported();
}
*out_obj = reinterpret_cast<void*>(new TestItem((char*)buf, size));
*charge = size;
return Status::OK();
};
void SetFailCreate(bool fail) { fail_create_ = fail; }
void BasicTestHelper(std::shared_ptr<SecondaryCache> sec_cache) {
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
bool is_in_sec_cache{true};
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
// Lookup an non-existent key.
std::unique_ptr<SecondaryCacheResultHandle> handle0 =
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
sec_cache->Lookup("k0", test_item_creator, true, is_in_sec_cache);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
ASSERT_EQ(handle0, nullptr);
Random rnd(301);
// Insert and Lookup the first item.
std::string str1;
test::CompressibleString(&rnd, 0.25, 1000, &str1);
TestItem item1(str1.data(), str1.length());
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_OK(sec_cache->Insert("k1", &item1,
&CompressedSecondaryCacheTest::helper_));
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
std::unique_ptr<SecondaryCacheResultHandle> handle1 =
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
sec_cache->Lookup("k1", test_item_creator, true, is_in_sec_cache);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
ASSERT_NE(handle1, nullptr);
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_FALSE(is_in_sec_cache);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
std::unique_ptr<TestItem> val1 =
std::unique_ptr<TestItem>(static_cast<TestItem*>(handle1->Value()));
ASSERT_NE(val1, nullptr);
ASSERT_EQ(memcmp(val1->Buf(), item1.Buf(), item1.Size()), 0);
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
// Lookup the first item again.
std::unique_ptr<SecondaryCacheResultHandle> handle1_1 =
sec_cache->Lookup("k1", test_item_creator, true, is_in_sec_cache);
ASSERT_EQ(handle1_1, nullptr);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
// Insert and Lookup the second item.
std::string str2;
test::CompressibleString(&rnd, 0.5, 1000, &str2);
TestItem item2(str2.data(), str2.length());
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_OK(sec_cache->Insert("k2", &item2,
&CompressedSecondaryCacheTest::helper_));
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
std::unique_ptr<SecondaryCacheResultHandle> handle2 =
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
sec_cache->Lookup("k2", test_item_creator, true, is_in_sec_cache);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
ASSERT_NE(handle2, nullptr);
std::unique_ptr<TestItem> val2 =
std::unique_ptr<TestItem>(static_cast<TestItem*>(handle2->Value()));
ASSERT_NE(val2, nullptr);
ASSERT_EQ(memcmp(val2->Buf(), item2.Buf(), item2.Size()), 0);
std::vector<SecondaryCacheResultHandle*> handles = {handle1.get(),
handle2.get()};
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
sec_cache->WaitAll(handles);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
sec_cache.reset();
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
}
void BasicTest(bool sec_cache_is_compressed, bool use_jemalloc) {
CompressedSecondaryCacheOptions opts;
opts.capacity = 2048;
opts.num_shard_bits = 0;
opts.metadata_charge_policy = kDontChargeCacheMetadata;
if (sec_cache_is_compressed) {
if (!LZ4_Supported()) {
ROCKSDB_GTEST_SKIP("This test requires LZ4 support.");
opts.compression_type = CompressionType::kNoCompression;
}
} else {
opts.compression_type = CompressionType::kNoCompression;
}
if (use_jemalloc) {
JemallocAllocatorOptions jopts;
std::shared_ptr<MemoryAllocator> allocator;
std::string msg;
if (JemallocNodumpAllocator::IsSupported(&msg)) {
Status s = NewJemallocNodumpAllocator(jopts, &allocator);
if (s.ok()) {
opts.memory_allocator = allocator;
}
} else {
ROCKSDB_GTEST_BYPASS("JEMALLOC not supported");
}
}
std::shared_ptr<SecondaryCache> sec_cache =
NewCompressedSecondaryCache(opts);
}
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
void FailsTest(bool sec_cache_is_compressed) {
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
CompressedSecondaryCacheOptions secondary_cache_opts;
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
if (sec_cache_is_compressed) {
if (!LZ4_Supported()) {
ROCKSDB_GTEST_SKIP("This test requires LZ4 support.");
secondary_cache_opts.compression_type = CompressionType::kNoCompression;
}
} else {
secondary_cache_opts.compression_type = CompressionType::kNoCompression;
}
secondary_cache_opts.capacity = 1100;
secondary_cache_opts.num_shard_bits = 0;
secondary_cache_opts.metadata_charge_policy = kDontChargeCacheMetadata;
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
std::shared_ptr<SecondaryCache> sec_cache =
NewCompressedSecondaryCache(secondary_cache_opts);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
// Insert and Lookup the first item.
Random rnd(301);
std::string str1(rnd.RandomString(1000));
TestItem item1(str1.data(), str1.length());
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_OK(sec_cache->Insert("k1", &item1,
&CompressedSecondaryCacheTest::helper_));
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
// Insert and Lookup the second item.
std::string str2(rnd.RandomString(200));
TestItem item2(str2.data(), str2.length());
// k1 is evicted.
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_OK(sec_cache->Insert("k2", &item2,
&CompressedSecondaryCacheTest::helper_));
bool is_in_sec_cache{false};
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
std::unique_ptr<SecondaryCacheResultHandle> handle1_1 =
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
sec_cache->Lookup("k1", test_item_creator, true, is_in_sec_cache);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
ASSERT_EQ(handle1_1, nullptr);
std::unique_ptr<SecondaryCacheResultHandle> handle2 =
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
sec_cache->Lookup("k2", test_item_creator, true, is_in_sec_cache);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
ASSERT_NE(handle2, nullptr);
std::unique_ptr<TestItem> val2 =
std::unique_ptr<TestItem>(static_cast<TestItem*>(handle2->Value()));
ASSERT_NE(val2, nullptr);
ASSERT_EQ(memcmp(val2->Buf(), item2.Buf(), item2.Size()), 0);
// Create Fails.
SetFailCreate(true);
std::unique_ptr<SecondaryCacheResultHandle> handle2_1 =
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
sec_cache->Lookup("k2", test_item_creator, true, is_in_sec_cache);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
ASSERT_EQ(handle2_1, nullptr);
// Save Fails.
std::string str3 = rnd.RandomString(10);
TestItem item3(str3.data(), str3.length());
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_NOK(sec_cache->Insert("k3", &item3,
&CompressedSecondaryCacheTest::helper_fail_));
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
sec_cache.reset();
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
}
void BasicIntegrationTest(bool sec_cache_is_compressed) {
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
CompressedSecondaryCacheOptions secondary_cache_opts;
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
if (sec_cache_is_compressed) {
if (!LZ4_Supported()) {
ROCKSDB_GTEST_SKIP("This test requires LZ4 support.");
secondary_cache_opts.compression_type = CompressionType::kNoCompression;
}
} else {
secondary_cache_opts.compression_type = CompressionType::kNoCompression;
}
secondary_cache_opts.capacity = 2300;
secondary_cache_opts.num_shard_bits = 0;
secondary_cache_opts.metadata_charge_policy = kDontChargeCacheMetadata;
std::shared_ptr<SecondaryCache> secondary_cache =
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
NewCompressedSecondaryCache(secondary_cache_opts);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
LRUCacheOptions lru_cache_opts(1024, 0, false, 0.5, nullptr,
kDefaultToAdaptiveMutex,
kDontChargeCacheMetadata);
lru_cache_opts.secondary_cache = secondary_cache;
std::shared_ptr<Cache> cache = NewLRUCache(lru_cache_opts);
std::shared_ptr<Statistics> stats = CreateDBStatistics();
Random rnd(301);
std::string str1 = rnd.RandomString(1010);
std::string str1_clone{str1};
TestItem* item1 = new TestItem(str1.data(), str1.length());
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_OK(cache->Insert("k1", item1, &CompressedSecondaryCacheTest::helper_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
str1.length()));
std::string str2 = rnd.RandomString(1020);
TestItem* item2 = new TestItem(str2.data(), str2.length());
// After Insert, lru cache contains k2 and secondary cache contains k1.
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_OK(cache->Insert("k2", item2, &CompressedSecondaryCacheTest::helper_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
str2.length()));
std::string str3 = rnd.RandomString(1020);
TestItem* item3 = new TestItem(str3.data(), str3.length());
// After Insert, lru cache contains k3 and secondary cache contains k1 and
// k2
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_OK(cache->Insert("k3", item3, &CompressedSecondaryCacheTest::helper_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
str3.length()));
Cache::Handle* handle;
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
handle = cache->Lookup("k3", &CompressedSecondaryCacheTest::helper_,
test_item_creator, Cache::Priority::LOW, true,
stats.get());
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
ASSERT_NE(handle, nullptr);
TestItem* val3 = static_cast<TestItem*>(cache->Value(handle));
ASSERT_NE(val3, nullptr);
ASSERT_EQ(memcmp(val3->Buf(), item3->Buf(), item3->Size()), 0);
cache->Release(handle);
// Lookup an non-existent key.
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
handle = cache->Lookup("k0", &CompressedSecondaryCacheTest::helper_,
test_item_creator, Cache::Priority::LOW, true,
stats.get());
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
ASSERT_EQ(handle, nullptr);
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
// This Lookup should promote k1 and erase k1 from the secondary cache,
// then k3 is demoted. So k2 and k3 are in the secondary cache.
handle = cache->Lookup("k1", &CompressedSecondaryCacheTest::helper_,
test_item_creator, Cache::Priority::LOW, true,
stats.get());
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
ASSERT_NE(handle, nullptr);
TestItem* val1_1 = static_cast<TestItem*>(cache->Value(handle));
ASSERT_NE(val1_1, nullptr);
ASSERT_EQ(memcmp(val1_1->Buf(), str1_clone.data(), str1_clone.size()), 0);
cache->Release(handle);
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
handle = cache->Lookup("k2", &CompressedSecondaryCacheTest::helper_,
test_item_creator, Cache::Priority::LOW, true,
stats.get());
ASSERT_NE(handle, nullptr);
cache->Release(handle);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
cache.reset();
secondary_cache.reset();
}
void BasicIntegrationFailTest(bool sec_cache_is_compressed) {
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
CompressedSecondaryCacheOptions secondary_cache_opts;
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
if (sec_cache_is_compressed) {
if (!LZ4_Supported()) {
ROCKSDB_GTEST_SKIP("This test requires LZ4 support.");
secondary_cache_opts.compression_type = CompressionType::kNoCompression;
}
} else {
secondary_cache_opts.compression_type = CompressionType::kNoCompression;
}
secondary_cache_opts.capacity = 2048;
secondary_cache_opts.num_shard_bits = 0;
secondary_cache_opts.metadata_charge_policy = kDontChargeCacheMetadata;
std::shared_ptr<SecondaryCache> secondary_cache =
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
NewCompressedSecondaryCache(secondary_cache_opts);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
LRUCacheOptions opts(1024, 0, false, 0.5, nullptr, kDefaultToAdaptiveMutex,
kDontChargeCacheMetadata);
opts.secondary_cache = secondary_cache;
std::shared_ptr<Cache> cache = NewLRUCache(opts);
Random rnd(301);
std::string str1 = rnd.RandomString(1020);
auto item1 =
std::unique_ptr<TestItem>(new TestItem(str1.data(), str1.length()));
ASSERT_NOK(cache->Insert("k1", item1.get(), nullptr, str1.length()));
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_OK(cache->Insert("k1", item1.get(),
&CompressedSecondaryCacheTest::helper_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
str1.length()));
item1.release(); // Appease clang-analyze "potential memory leak"
Cache::Handle* handle;
handle = cache->Lookup("k2", nullptr, test_item_creator,
Cache::Priority::LOW, true);
ASSERT_EQ(handle, nullptr);
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
handle = cache->Lookup("k2", &CompressedSecondaryCacheTest::helper_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
test_item_creator, Cache::Priority::LOW, false);
ASSERT_EQ(handle, nullptr);
cache.reset();
secondary_cache.reset();
}
void IntegrationSaveFailTest(bool sec_cache_is_compressed) {
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
CompressedSecondaryCacheOptions secondary_cache_opts;
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
if (sec_cache_is_compressed) {
if (!LZ4_Supported()) {
ROCKSDB_GTEST_SKIP("This test requires LZ4 support.");
secondary_cache_opts.compression_type = CompressionType::kNoCompression;
}
} else {
secondary_cache_opts.compression_type = CompressionType::kNoCompression;
}
secondary_cache_opts.capacity = 2048;
secondary_cache_opts.num_shard_bits = 0;
secondary_cache_opts.metadata_charge_policy = kDontChargeCacheMetadata;
std::shared_ptr<SecondaryCache> secondary_cache =
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
NewCompressedSecondaryCache(secondary_cache_opts);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
LRUCacheOptions opts(1024, 0, false, 0.5, nullptr, kDefaultToAdaptiveMutex,
kDontChargeCacheMetadata);
opts.secondary_cache = secondary_cache;
std::shared_ptr<Cache> cache = NewLRUCache(opts);
Random rnd(301);
std::string str1 = rnd.RandomString(1020);
TestItem* item1 = new TestItem(str1.data(), str1.length());
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_OK(cache->Insert("k1", item1,
&CompressedSecondaryCacheTest::helper_fail_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
str1.length()));
std::string str2 = rnd.RandomString(1020);
TestItem* item2 = new TestItem(str2.data(), str2.length());
// k1 should be demoted to the secondary cache.
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_OK(cache->Insert("k2", item2,
&CompressedSecondaryCacheTest::helper_fail_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
str2.length()));
Cache::Handle* handle;
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
handle = cache->Lookup("k2", &CompressedSecondaryCacheTest::helper_fail_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
test_item_creator, Cache::Priority::LOW, true);
ASSERT_NE(handle, nullptr);
cache->Release(handle);
// This lookup should fail, since k1 demotion would have failed
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
handle = cache->Lookup("k1", &CompressedSecondaryCacheTest::helper_fail_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
test_item_creator, Cache::Priority::LOW, true);
ASSERT_EQ(handle, nullptr);
// Since k1 didn't get promoted, k2 should still be in cache
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
handle = cache->Lookup("k2", &CompressedSecondaryCacheTest::helper_fail_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
test_item_creator, Cache::Priority::LOW, true);
ASSERT_NE(handle, nullptr);
cache->Release(handle);
cache.reset();
secondary_cache.reset();
}
void IntegrationCreateFailTest(bool sec_cache_is_compressed) {
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
CompressedSecondaryCacheOptions secondary_cache_opts;
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
if (sec_cache_is_compressed) {
if (!LZ4_Supported()) {
ROCKSDB_GTEST_SKIP("This test requires LZ4 support.");
secondary_cache_opts.compression_type = CompressionType::kNoCompression;
}
} else {
secondary_cache_opts.compression_type = CompressionType::kNoCompression;
}
secondary_cache_opts.capacity = 2048;
secondary_cache_opts.num_shard_bits = 0;
secondary_cache_opts.metadata_charge_policy = kDontChargeCacheMetadata;
std::shared_ptr<SecondaryCache> secondary_cache =
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
NewCompressedSecondaryCache(secondary_cache_opts);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
LRUCacheOptions opts(1024, 0, false, 0.5, nullptr, kDefaultToAdaptiveMutex,
kDontChargeCacheMetadata);
opts.secondary_cache = secondary_cache;
std::shared_ptr<Cache> cache = NewLRUCache(opts);
Random rnd(301);
std::string str1 = rnd.RandomString(1020);
TestItem* item1 = new TestItem(str1.data(), str1.length());
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_OK(cache->Insert("k1", item1, &CompressedSecondaryCacheTest::helper_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
str1.length()));
std::string str2 = rnd.RandomString(1020);
TestItem* item2 = new TestItem(str2.data(), str2.length());
// k1 should be demoted to the secondary cache.
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_OK(cache->Insert("k2", item2, &CompressedSecondaryCacheTest::helper_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
str2.length()));
Cache::Handle* handle;
SetFailCreate(true);
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
handle = cache->Lookup("k2", &CompressedSecondaryCacheTest::helper_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
test_item_creator, Cache::Priority::LOW, true);
ASSERT_NE(handle, nullptr);
cache->Release(handle);
// This lookup should fail, since k1 creation would have failed
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
handle = cache->Lookup("k1", &CompressedSecondaryCacheTest::helper_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
test_item_creator, Cache::Priority::LOW, true);
ASSERT_EQ(handle, nullptr);
// Since k1 didn't get promoted, k2 should still be in cache
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
handle = cache->Lookup("k2", &CompressedSecondaryCacheTest::helper_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
test_item_creator, Cache::Priority::LOW, true);
ASSERT_NE(handle, nullptr);
cache->Release(handle);
cache.reset();
secondary_cache.reset();
}
void IntegrationFullCapacityTest(bool sec_cache_is_compressed) {
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
CompressedSecondaryCacheOptions secondary_cache_opts;
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
if (sec_cache_is_compressed) {
if (!LZ4_Supported()) {
ROCKSDB_GTEST_SKIP("This test requires LZ4 support.");
secondary_cache_opts.compression_type = CompressionType::kNoCompression;
}
} else {
secondary_cache_opts.compression_type = CompressionType::kNoCompression;
}
secondary_cache_opts.capacity = 2048;
secondary_cache_opts.num_shard_bits = 0;
secondary_cache_opts.metadata_charge_policy = kDontChargeCacheMetadata;
std::shared_ptr<SecondaryCache> secondary_cache =
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
NewCompressedSecondaryCache(secondary_cache_opts);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
LRUCacheOptions opts(1024, 0, /*_strict_capacity_limit=*/true, 0.5, nullptr,
kDefaultToAdaptiveMutex, kDontChargeCacheMetadata);
opts.secondary_cache = secondary_cache;
std::shared_ptr<Cache> cache = NewLRUCache(opts);
Random rnd(301);
std::string str1 = rnd.RandomString(1020);
TestItem* item1 = new TestItem(str1.data(), str1.length());
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_OK(cache->Insert("k1", item1, &CompressedSecondaryCacheTest::helper_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
str1.length()));
std::string str2 = rnd.RandomString(1020);
TestItem* item2 = new TestItem(str2.data(), str2.length());
// k1 should be demoted to the secondary cache.
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
ASSERT_OK(cache->Insert("k2", item2, &CompressedSecondaryCacheTest::helper_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
str2.length()));
Cache::Handle* handle2;
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
handle2 = cache->Lookup("k2", &CompressedSecondaryCacheTest::helper_,
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
test_item_creator, Cache::Priority::LOW, true);
ASSERT_NE(handle2, nullptr);
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
cache->Release(handle2);
// k1 promotion should fail due to the block cache being at capacity,
// but the lookup should still succeed
Cache::Handle* handle1;
handle1 = cache->Lookup("k1", &CompressedSecondaryCacheTest::helper_,
test_item_creator, Cache::Priority::LOW, true);
ASSERT_NE(handle1, nullptr);
cache->Release(handle1);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
// Since k1 didn't get inserted, k2 should still be in cache
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
handle2 = cache->Lookup("k2", &CompressedSecondaryCacheTest::helper_,
test_item_creator, Cache::Priority::LOW, true);
ASSERT_NE(handle2, nullptr);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
cache->Release(handle2);
cache.reset();
secondary_cache.reset();
}
private:
bool fail_create_;
};
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
Cache::CacheItemHelper CompressedSecondaryCacheTest::helper_(
CompressedSecondaryCacheTest::SizeCallback,
CompressedSecondaryCacheTest::SaveToCallback,
CompressedSecondaryCacheTest::DeletionCallback);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
Cache::CacheItemHelper CompressedSecondaryCacheTest::helper_fail_(
CompressedSecondaryCacheTest::SizeCallback,
CompressedSecondaryCacheTest::SaveToCallbackFail,
CompressedSecondaryCacheTest::DeletionCallback);
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest, BasicTestWithNoCompression) {
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
BasicTest(false, false);
}
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest,
BasicTestWithMemoryAllocatorAndNoCompression) {
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
BasicTest(false, true);
}
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest, BasicTestWithCompression) {
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
BasicTest(true, false);
}
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest,
BasicTestWithMemoryAllocatorAndCompression) {
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
BasicTest(true, true);
}
#ifndef ROCKSDB_LITE
TEST_F(CompressedSecondaryCacheTest, BasicTestFromStringWithNoCompression) {
std::string sec_cache_uri =
"compressed_secondary_cache://"
"capacity=2048;num_shard_bits=0;compression_type=kNoCompression";
std::shared_ptr<SecondaryCache> sec_cache;
Status s = SecondaryCache::CreateFromString(ConfigOptions(), sec_cache_uri,
&sec_cache);
EXPECT_OK(s);
BasicTestHelper(sec_cache);
}
TEST_F(CompressedSecondaryCacheTest, BasicTestFromStringWithCompression) {
std::string sec_cache_uri;
if (LZ4_Supported()) {
sec_cache_uri =
"compressed_secondary_cache://"
"capacity=2048;num_shard_bits=0;compression_type=kLZ4Compression;"
"compress_format_version=2";
} else {
ROCKSDB_GTEST_SKIP("This test requires LZ4 support.");
sec_cache_uri =
"compressed_secondary_cache://"
"capacity=2048;num_shard_bits=0;compression_type=kNoCompression";
}
std::shared_ptr<SecondaryCache> sec_cache;
Status s = SecondaryCache::CreateFromString(ConfigOptions(), sec_cache_uri,
&sec_cache);
EXPECT_OK(s);
BasicTestHelper(sec_cache);
}
#endif // ROCKSDB_LITE
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest, FailsTestWithNoCompression) {
FailsTest(false);
}
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest, FailsTestWithCompression) {
FailsTest(true);
}
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest, BasicIntegrationTestWithNoCompression) {
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
BasicIntegrationTest(false);
}
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest, BasicIntegrationTestWithCompression) {
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
BasicIntegrationTest(true);
}
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest,
BasicIntegrationFailTestWithNoCompression) {
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
BasicIntegrationFailTest(false);
}
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest, BasicIntegrationFailTestWithCompression) {
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
BasicIntegrationFailTest(true);
}
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest, IntegrationSaveFailTestWithNoCompression) {
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
IntegrationSaveFailTest(false);
}
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest, IntegrationSaveFailTestWithCompression) {
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
IntegrationSaveFailTest(true);
}
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest,
IntegrationCreateFailTestWithNoCompression) {
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
IntegrationCreateFailTest(false);
}
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest, IntegrationCreateFailTestWithCompression) {
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
IntegrationCreateFailTest(true);
}
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest,
IntegrationFullCapacityTestWithNoCompression) {
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
IntegrationFullCapacityTest(false);
}
Prevent double caching in the compressed secondary cache (#9747) Summary: ### **Summary:** When both LRU Cache and CompressedSecondaryCache are configured together, there possibly are some data blocks double cached. **Changes include:** 1. Update IS_PROMOTED to IS_IN_SECONDARY_CACHE to prevent confusions. 2. This PR updates SecondaryCacheResultHandle and use IsErasedFromSecondaryCache to determine whether the handle is erased in the secondary cache. Then, the caller can determine whether to SetIsInSecondaryCache(). 3. Rename LRUSecondaryCache to CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9747 Test Plan: **Test Scripts:** 1. Populate a DB. The on disk footprint is 482 MB. The data is set to be 50% compressible, so the total decompressed size is expected to be 964 MB. ./db_bench --benchmarks=fillrandom --num=10000000 -db=/db_bench_1 2. overwrite it to a stable state: ./db_bench --benchmarks=overwrite,stats --num=10000000 -use_existing_db -duration=10 --benchmark_write_rate_limit=2000000 -db=/db_bench_1 4. Run read tests with diffeernt cache setting: T1: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 --statistics -db=/db_bench_1 T2: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=320000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T3: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=520000000 -compressed_secondary_cache_size=400000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 T4: ./db_bench --benchmarks=seekrandom,stats --threads=16 --num=10000000 -use_existing_db -duration=120 --benchmark_write_rate_limit=52000000 -use_direct_reads --cache_size=20000000 -compressed_secondary_cache_size=500000000 --statistics -use_compressed_secondary_cache -db=/db_bench_1 **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 96.2% | |520 MB | 400 MB | 98.3% | |20 MB | 500 MB | 98.8% | **Before this PR** | Cache Size | Compressed Secondary Cache Size | Cache Hit Rate | |------------|-------------------------------------|----------------| |520 MB | 0 MB | 85.5% | |320 MB | 400 MB | 99.9% | |520 MB | 400 MB | 99.9% | |20 MB | 500 MB | 99.2% | Reviewed By: anand1976 Differential Revision: D35117499 Pulled By: gitbw95 fbshipit-source-id: ea2657749fc13efebe91a8a1b56bc61d6a224a12
3 years ago
TEST_F(CompressedSecondaryCacheTest,
IntegrationFullCapacityTestWithCompression) {
Add a secondary cache implementation based on LRUCache 1 (#9518) Summary: **Summary:** RocksDB uses a block cache to reduce IO and make queries more efficient. The block cache is based on the LRU algorithm (LRUCache) and keeps objects containing uncompressed data, such as Block, ParsedFullFilterBlock etc. It allows the user to configure a second level cache (rocksdb::SecondaryCache) to extend the primary block cache by holding items evicted from it. Some of the major RocksDB users, like MyRocks, use direct IO and would like to use a primary block cache for uncompressed data and a secondary cache for compressed data. The latter allows us to mitigate the loss of the Linux page cache due to direct IO. This PR includes a concrete implementation of rocksdb::SecondaryCache that integrates with compression libraries such as LZ4 and implements an LRU cache to hold compressed blocks. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9518 Test Plan: In this PR, the lru_secondary_cache_test.cc includes the following tests: 1. The unit tests for the secondary cache with either compression or no compression, such as basic tests, fails tests. 2. The integration tests with both primary cache and this secondary cache . **Follow Up:** 1. Statistics (e.g. compression ratio) will be added in another PR. 2. Once this implementation is ready, I will do some shadow testing and benchmarking with UDB to measure the impact. Reviewed By: anand1976 Differential Revision: D34430930 Pulled By: gitbw95 fbshipit-source-id: 218d78b672a2f914856d8a90ff32f2f5b5043ded
3 years ago
IntegrationFullCapacityTest(true);
}
} // namespace ROCKSDB_NAMESPACE
int main(int argc, char** argv) {
::testing::InitGoogleTest(&argc, argv);
return RUN_ALL_TESTS();
}