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/table/block_based/partitioned_filter_block_te...

424 lines
15 KiB

// 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).
#include <map>
#include "rocksdb/filter_policy.h"
#include "table/block_based/block_based_table_reader.h"
#include "table/block_based/partitioned_filter_block.h"
#include "table/block_based/filter_policy_internal.h"
#include "index_builder.h"
#include "logging/logging.h"
#include "test_util/testharness.h"
#include "test_util/testutil.h"
#include "util/coding.h"
#include "util/hash.h"
namespace rocksdb {
Store the filter bits reader alongside the filter block contents (#5936) Summary: Amongst other things, PR https://github.com/facebook/rocksdb/issues/5504 refactored the filter block readers so that only the filter block contents are stored in the block cache (as opposed to the earlier design where the cache stored the filter block reader itself, leading to potentially dangling pointers and concurrency bugs). However, this change introduced a performance hit since with the new code, the metadata fields are re-parsed upon every access. This patch reunites the block contents with the filter bits reader to eliminate this overhead; since this is still a self-contained pure data object, it is safe to store it in the cache. (Note: this is similar to how the zstd digest is handled.) Pull Request resolved: https://github.com/facebook/rocksdb/pull/5936 Test Plan: make asan_check filter_bench results for the old code: ``` $ ./filter_bench -quick WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 26.7153 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 33.4258 Single filter ns/op: 42.5974 Random filter ns/op: 217.861 ---------------------------- Outside queries... Dry run (25d) ns/op: 32.4217 Single filter ns/op: 50.9855 Random filter ns/op: 219.167 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) $ ./filter_bench -quick -use_full_block_reader WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 26.5172 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 32.3556 Single filter ns/op: 83.2239 Random filter ns/op: 370.676 ---------------------------- Outside queries... Dry run (25d) ns/op: 32.2265 Single filter ns/op: 93.5651 Random filter ns/op: 408.393 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) ``` With the new code: ``` $ ./filter_bench -quick WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 25.4285 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 31.0594 Single filter ns/op: 43.8974 Random filter ns/op: 226.075 ---------------------------- Outside queries... Dry run (25d) ns/op: 31.0295 Single filter ns/op: 50.3824 Random filter ns/op: 226.805 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) $ ./filter_bench -quick -use_full_block_reader WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 26.5308 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 33.2968 Single filter ns/op: 58.6163 Random filter ns/op: 291.434 ---------------------------- Outside queries... Dry run (25d) ns/op: 32.1839 Single filter ns/op: 66.9039 Random filter ns/op: 292.828 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) ``` Differential Revision: D17991712 Pulled By: ltamasi fbshipit-source-id: 7ea205550217bfaaa1d5158ebd658e5832e60f29
5 years ago
std::map<uint64_t, std::string> blooms;
class MockedBlockBasedTable : public BlockBasedTable {
public:
MockedBlockBasedTable(Rep* rep, PartitionedIndexBuilder* pib)
: BlockBasedTable(rep, /*block_cache_tracer=*/nullptr) {
// Initialize what Open normally does as much as necessary for the test
rep->index_key_includes_seq = pib->seperator_is_key_plus_seq();
rep->index_value_is_full = !pib->get_use_value_delta_encoding();
}
};
class MyPartitionedFilterBlockReader : public PartitionedFilterBlockReader {
public:
MyPartitionedFilterBlockReader(BlockBasedTable* t,
CachableEntry<Block>&& filter_block)
: PartitionedFilterBlockReader(t, std::move(filter_block)) {
Store the filter bits reader alongside the filter block contents (#5936) Summary: Amongst other things, PR https://github.com/facebook/rocksdb/issues/5504 refactored the filter block readers so that only the filter block contents are stored in the block cache (as opposed to the earlier design where the cache stored the filter block reader itself, leading to potentially dangling pointers and concurrency bugs). However, this change introduced a performance hit since with the new code, the metadata fields are re-parsed upon every access. This patch reunites the block contents with the filter bits reader to eliminate this overhead; since this is still a self-contained pure data object, it is safe to store it in the cache. (Note: this is similar to how the zstd digest is handled.) Pull Request resolved: https://github.com/facebook/rocksdb/pull/5936 Test Plan: make asan_check filter_bench results for the old code: ``` $ ./filter_bench -quick WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 26.7153 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 33.4258 Single filter ns/op: 42.5974 Random filter ns/op: 217.861 ---------------------------- Outside queries... Dry run (25d) ns/op: 32.4217 Single filter ns/op: 50.9855 Random filter ns/op: 219.167 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) $ ./filter_bench -quick -use_full_block_reader WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 26.5172 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 32.3556 Single filter ns/op: 83.2239 Random filter ns/op: 370.676 ---------------------------- Outside queries... Dry run (25d) ns/op: 32.2265 Single filter ns/op: 93.5651 Random filter ns/op: 408.393 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) ``` With the new code: ``` $ ./filter_bench -quick WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 25.4285 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 31.0594 Single filter ns/op: 43.8974 Random filter ns/op: 226.075 ---------------------------- Outside queries... Dry run (25d) ns/op: 31.0295 Single filter ns/op: 50.3824 Random filter ns/op: 226.805 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) $ ./filter_bench -quick -use_full_block_reader WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 26.5308 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 33.2968 Single filter ns/op: 58.6163 Random filter ns/op: 291.434 ---------------------------- Outside queries... Dry run (25d) ns/op: 32.1839 Single filter ns/op: 66.9039 Random filter ns/op: 292.828 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) ``` Differential Revision: D17991712 Pulled By: ltamasi fbshipit-source-id: 7ea205550217bfaaa1d5158ebd658e5832e60f29
5 years ago
for (const auto& pair : blooms) {
const uint64_t offset = pair.first;
Store the filter bits reader alongside the filter block contents (#5936) Summary: Amongst other things, PR https://github.com/facebook/rocksdb/issues/5504 refactored the filter block readers so that only the filter block contents are stored in the block cache (as opposed to the earlier design where the cache stored the filter block reader itself, leading to potentially dangling pointers and concurrency bugs). However, this change introduced a performance hit since with the new code, the metadata fields are re-parsed upon every access. This patch reunites the block contents with the filter bits reader to eliminate this overhead; since this is still a self-contained pure data object, it is safe to store it in the cache. (Note: this is similar to how the zstd digest is handled.) Pull Request resolved: https://github.com/facebook/rocksdb/pull/5936 Test Plan: make asan_check filter_bench results for the old code: ``` $ ./filter_bench -quick WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 26.7153 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 33.4258 Single filter ns/op: 42.5974 Random filter ns/op: 217.861 ---------------------------- Outside queries... Dry run (25d) ns/op: 32.4217 Single filter ns/op: 50.9855 Random filter ns/op: 219.167 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) $ ./filter_bench -quick -use_full_block_reader WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 26.5172 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 32.3556 Single filter ns/op: 83.2239 Random filter ns/op: 370.676 ---------------------------- Outside queries... Dry run (25d) ns/op: 32.2265 Single filter ns/op: 93.5651 Random filter ns/op: 408.393 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) ``` With the new code: ``` $ ./filter_bench -quick WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 25.4285 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 31.0594 Single filter ns/op: 43.8974 Random filter ns/op: 226.075 ---------------------------- Outside queries... Dry run (25d) ns/op: 31.0295 Single filter ns/op: 50.3824 Random filter ns/op: 226.805 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) $ ./filter_bench -quick -use_full_block_reader WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 26.5308 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 33.2968 Single filter ns/op: 58.6163 Random filter ns/op: 291.434 ---------------------------- Outside queries... Dry run (25d) ns/op: 32.1839 Single filter ns/op: 66.9039 Random filter ns/op: 292.828 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) ``` Differential Revision: D17991712 Pulled By: ltamasi fbshipit-source-id: 7ea205550217bfaaa1d5158ebd658e5832e60f29
5 years ago
const std::string& bloom = pair.second;
assert(t);
assert(t->get_rep());
CachableEntry<ParsedFullFilterBlock> block(
new ParsedFullFilterBlock(
t->get_rep()->table_options.filter_policy.get(),
BlockContents(Slice(bloom))),
nullptr /* cache */, nullptr /* cache_handle */,
true /* own_value */);
filter_map_[offset] = std::move(block);
}
}
};
class PartitionedFilterBlockTest
: public testing::Test,
virtual public ::testing::WithParamInterface<uint32_t> {
public:
Options options_;
ImmutableCFOptions ioptions_;
EnvOptions env_options_;
BlockBasedTableOptions table_options_;
InternalKeyComparator icomp_;
std::unique_ptr<BlockBasedTable> table_;
std::shared_ptr<Cache> cache_;
int bits_per_key_;
PartitionedFilterBlockTest()
: ioptions_(options_),
env_options_(options_),
icomp_(options_.comparator),
bits_per_key_(10) {
table_options_.filter_policy.reset(
NewBloomFilterPolicy(bits_per_key_, false));
table_options_.format_version = GetParam();
table_options_.index_block_restart_interval = 3;
}
~PartitionedFilterBlockTest() override {}
const std::string keys[4] = {"afoo", "bar", "box", "hello"};
const std::string missing_keys[2] = {"missing", "other"};
uint64_t MaxIndexSize() {
int num_keys = sizeof(keys) / sizeof(*keys);
uint64_t max_key_size = 0;
for (int i = 1; i < num_keys; i++) {
max_key_size = std::max(max_key_size, static_cast<uint64_t>(keys[i].size()));
}
uint64_t max_index_size = num_keys * (max_key_size + 8 /*handle*/);
return max_index_size;
}
uint64_t MaxFilterSize() {
int num_keys = sizeof(keys) / sizeof(*keys);
// General, rough over-approximation
return num_keys * bits_per_key_ + (CACHE_LINE_SIZE * 8 + /*metadata*/ 5);
}
uint64_t last_offset = 10;
BlockHandle Write(const Slice& slice) {
BlockHandle bh(last_offset + 1, slice.size());
Store the filter bits reader alongside the filter block contents (#5936) Summary: Amongst other things, PR https://github.com/facebook/rocksdb/issues/5504 refactored the filter block readers so that only the filter block contents are stored in the block cache (as opposed to the earlier design where the cache stored the filter block reader itself, leading to potentially dangling pointers and concurrency bugs). However, this change introduced a performance hit since with the new code, the metadata fields are re-parsed upon every access. This patch reunites the block contents with the filter bits reader to eliminate this overhead; since this is still a self-contained pure data object, it is safe to store it in the cache. (Note: this is similar to how the zstd digest is handled.) Pull Request resolved: https://github.com/facebook/rocksdb/pull/5936 Test Plan: make asan_check filter_bench results for the old code: ``` $ ./filter_bench -quick WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 26.7153 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 33.4258 Single filter ns/op: 42.5974 Random filter ns/op: 217.861 ---------------------------- Outside queries... Dry run (25d) ns/op: 32.4217 Single filter ns/op: 50.9855 Random filter ns/op: 219.167 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) $ ./filter_bench -quick -use_full_block_reader WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 26.5172 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 32.3556 Single filter ns/op: 83.2239 Random filter ns/op: 370.676 ---------------------------- Outside queries... Dry run (25d) ns/op: 32.2265 Single filter ns/op: 93.5651 Random filter ns/op: 408.393 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) ``` With the new code: ``` $ ./filter_bench -quick WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 25.4285 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 31.0594 Single filter ns/op: 43.8974 Random filter ns/op: 226.075 ---------------------------- Outside queries... Dry run (25d) ns/op: 31.0295 Single filter ns/op: 50.3824 Random filter ns/op: 226.805 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) $ ./filter_bench -quick -use_full_block_reader WARNING: Assertions are enabled; benchmarks unnecessarily slow Building... Build avg ns/key: 26.5308 Number of filters: 16669 Total memory (MB): 200.009 Bits/key actual: 10.0647 ---------------------------- Inside queries... Dry run (46b) ns/op: 33.2968 Single filter ns/op: 58.6163 Random filter ns/op: 291.434 ---------------------------- Outside queries... Dry run (25d) ns/op: 32.1839 Single filter ns/op: 66.9039 Random filter ns/op: 292.828 Average FP rate %: 1.13993 ---------------------------- Done. (For more info, run with -legend or -help.) ``` Differential Revision: D17991712 Pulled By: ltamasi fbshipit-source-id: 7ea205550217bfaaa1d5158ebd658e5832e60f29
5 years ago
blooms[bh.offset()] = slice.ToString();
last_offset += bh.size();
return bh;
}
PartitionedIndexBuilder* NewIndexBuilder() {
const bool kValueDeltaEncoded = true;
return PartitionedIndexBuilder::CreateIndexBuilder(
&icomp_, !kValueDeltaEncoded, table_options_);
}
PartitionedFilterBlockBuilder* NewBuilder(
PartitionedIndexBuilder* const p_index_builder,
const SliceTransform* prefix_extractor = nullptr) {
assert(table_options_.block_size_deviation <= 100);
auto partition_size = static_cast<uint32_t>(
((table_options_.metadata_block_size *
(100 - table_options_.block_size_deviation)) +
99) /
100);
partition_size = std::max(partition_size, static_cast<uint32_t>(1));
const bool kValueDeltaEncoded = true;
return new PartitionedFilterBlockBuilder(
prefix_extractor, table_options_.whole_key_filtering,
New Bloom filter implementation for full and partitioned filters (#6007) Summary: Adds an improved, replacement Bloom filter implementation (FastLocalBloom) for full and partitioned filters in the block-based table. This replacement is faster and more accurate, especially for high bits per key or millions of keys in a single filter. Speed The improved speed, at least on recent x86_64, comes from * Using fastrange instead of modulo (%) * Using our new hash function (XXH3 preview, added in a previous commit), which is much faster for large keys and only *slightly* slower on keys around 12 bytes if hashing the same size many thousands of times in a row. * Optimizing the Bloom filter queries with AVX2 SIMD operations. (Added AVX2 to the USE_SSE=1 build.) Careful design was required to support (a) SIMD-optimized queries, (b) compatible non-SIMD code that's simple and efficient, (c) flexible choice of number of probes, and (d) essentially maximized accuracy for a cache-local Bloom filter. Probes are made eight at a time, so any number of probes up to 8 is the same speed, then up to 16, etc. * Prefetching cache lines when building the filter. Although this optimization could be applied to the old structure as well, it seems to balance out the small added cost of accumulating 64 bit hashes for adding to the filter rather than 32 bit hashes. Here's nominal speed data from filter_bench (200MB in filters, about 10k keys each, 10 bits filter data / key, 6 probes, avg key size 24 bytes, includes hashing time) on Skylake DE (relatively low clock speed): $ ./filter_bench -quick -impl=2 -net_includes_hashing # New Bloom filter Build avg ns/key: 47.7135 Mixed inside/outside queries... Single filter net ns/op: 26.2825 Random filter net ns/op: 150.459 Average FP rate %: 0.954651 $ ./filter_bench -quick -impl=0 -net_includes_hashing # Old Bloom filter Build avg ns/key: 47.2245 Mixed inside/outside queries... Single filter net ns/op: 63.2978 Random filter net ns/op: 188.038 Average FP rate %: 1.13823 Similar build time but dramatically faster query times on hot data (63 ns to 26 ns), and somewhat faster on stale data (188 ns to 150 ns). Performance differences on batched and skewed query loads are between these extremes as expected. The only other interesting thing about speed is "inside" (query key was added to filter) vs. "outside" (query key was not added to filter) query times. The non-SIMD implementations are substantially slower when most queries are "outside" vs. "inside". This goes against what one might expect or would have observed years ago, as "outside" queries only need about two probes on average, due to short-circuiting, while "inside" always have num_probes (say 6). The problem is probably the nastily unpredictable branch. The SIMD implementation has few branches (very predictable) and has pretty consistent running time regardless of query outcome. Accuracy The generally improved accuracy (re: Issue https://github.com/facebook/rocksdb/issues/5857) comes from a better design for probing indices within a cache line (re: Issue https://github.com/facebook/rocksdb/issues/4120) and improved accuracy for millions of keys in a single filter from using a 64-bit hash function (XXH3p). Design details in code comments. Accuracy data (generalizes, except old impl gets worse with millions of keys): Memory bits per key: FP rate percent old impl -> FP rate percent new impl 6: 5.70953 -> 5.69888 8: 2.45766 -> 2.29709 10: 1.13977 -> 0.959254 12: 0.662498 -> 0.411593 16: 0.353023 -> 0.0873754 24: 0.261552 -> 0.0060971 50: 0.225453 -> ~0.00003 (less than 1 in a million queries are FP) Fixes https://github.com/facebook/rocksdb/issues/5857 Fixes https://github.com/facebook/rocksdb/issues/4120 Unlike the old implementation, this implementation has a fixed cache line size (64 bytes). At 10 bits per key, the accuracy of this new implementation is very close to the old implementation with 128-byte cache line size. If there's sufficient demand, this implementation could be generalized. Compatibility Although old releases would see the new structure as corrupt filter data and read the table as if there's no filter, we've decided only to enable the new Bloom filter with new format_version=5. This provides a smooth path for automatic adoption over time, with an option for early opt-in. Pull Request resolved: https://github.com/facebook/rocksdb/pull/6007 Test Plan: filter_bench has been used thoroughly to validate speed, accuracy, and correctness. Unit tests have been carefully updated to exercise new and old implementations, as well as the logic to select an implementation based on context (format_version). Differential Revision: D18294749 Pulled By: pdillinger fbshipit-source-id: d44c9db3696e4d0a17caaec47075b7755c262c5f
5 years ago
FilterBuildingContext(table_options_).GetBuilder(),
table_options_.index_block_restart_interval, !kValueDeltaEncoded,
p_index_builder, partition_size);
}
PartitionedFilterBlockReader* NewReader(
PartitionedFilterBlockBuilder* builder, PartitionedIndexBuilder* pib) {
BlockHandle bh;
Status status;
Slice slice;
do {
slice = builder->Finish(bh, &status);
bh = Write(slice);
} while (status.IsIncomplete());
constexpr bool skip_filters = false;
constexpr int level = 0;
constexpr bool immortal_table = false;
table_.reset(new MockedBlockBasedTable(
new BlockBasedTable::Rep(ioptions_, env_options_, table_options_,
icomp_, skip_filters, level, immortal_table),
pib));
BlockContents contents(slice);
CachableEntry<Block> block(
new Block(std::move(contents), kDisableGlobalSequenceNumber,
0 /* read_amp_bytes_per_bit */, nullptr),
nullptr /* cache */, nullptr /* cache_handle */, true /* own_value */);
auto reader =
new MyPartitionedFilterBlockReader(table_.get(), std::move(block));
return reader;
}
void VerifyReader(PartitionedFilterBlockBuilder* builder,
PartitionedIndexBuilder* pib, bool empty = false,
const SliceTransform* prefix_extractor = nullptr) {
std::unique_ptr<PartitionedFilterBlockReader> reader(
NewReader(builder, pib));
// Querying added keys
const bool no_io = true;
for (auto key : keys) {
auto ikey = InternalKey(key, 0, ValueType::kTypeValue);
const Slice ikey_slice = Slice(*ikey.rep());
ASSERT_TRUE(reader->KeyMayMatch(key, prefix_extractor, kNotValid, !no_io,
&ikey_slice, /*get_context=*/nullptr,
/*lookup_context=*/nullptr));
}
{
// querying a key twice
auto ikey = InternalKey(keys[0], 0, ValueType::kTypeValue);
const Slice ikey_slice = Slice(*ikey.rep());
ASSERT_TRUE(reader->KeyMayMatch(
keys[0], prefix_extractor, kNotValid, !no_io, &ikey_slice,
/*get_context=*/nullptr, /*lookup_context=*/nullptr));
}
// querying missing keys
for (auto key : missing_keys) {
auto ikey = InternalKey(key, 0, ValueType::kTypeValue);
const Slice ikey_slice = Slice(*ikey.rep());
if (empty) {
ASSERT_TRUE(reader->KeyMayMatch(
key, prefix_extractor, kNotValid, !no_io, &ikey_slice,
/*get_context=*/nullptr, /*lookup_context=*/nullptr));
} else {
// assuming a good hash function
ASSERT_FALSE(reader->KeyMayMatch(
key, prefix_extractor, kNotValid, !no_io, &ikey_slice,
/*get_context=*/nullptr, /*lookup_context=*/nullptr));
}
}
}
int TestBlockPerKey() {
std::unique_ptr<PartitionedIndexBuilder> pib(NewIndexBuilder());
std::unique_ptr<PartitionedFilterBlockBuilder> builder(
NewBuilder(pib.get()));
int i = 0;
builder->Add(keys[i]);
CutABlock(pib.get(), keys[i], keys[i + 1]);
i++;
builder->Add(keys[i]);
CutABlock(pib.get(), keys[i], keys[i + 1]);
i++;
builder->Add(keys[i]);
builder->Add(keys[i]);
CutABlock(pib.get(), keys[i], keys[i + 1]);
i++;
builder->Add(keys[i]);
CutABlock(pib.get(), keys[i]);
VerifyReader(builder.get(), pib.get());
return CountNumOfIndexPartitions(pib.get());
}
void TestBlockPerTwoKeys(const SliceTransform* prefix_extractor = nullptr) {
std::unique_ptr<PartitionedIndexBuilder> pib(NewIndexBuilder());
std::unique_ptr<PartitionedFilterBlockBuilder> builder(
NewBuilder(pib.get(), prefix_extractor));
int i = 0;
builder->Add(keys[i]);
i++;
builder->Add(keys[i]);
CutABlock(pib.get(), keys[i], keys[i + 1]);
i++;
builder->Add(keys[i]);
builder->Add(keys[i]);
i++;
builder->Add(keys[i]);
CutABlock(pib.get(), keys[i]);
VerifyReader(builder.get(), pib.get(), prefix_extractor);
}
void TestBlockPerAllKeys() {
std::unique_ptr<PartitionedIndexBuilder> pib(NewIndexBuilder());
std::unique_ptr<PartitionedFilterBlockBuilder> builder(
NewBuilder(pib.get()));
int i = 0;
builder->Add(keys[i]);
i++;
builder->Add(keys[i]);
i++;
builder->Add(keys[i]);
builder->Add(keys[i]);
i++;
builder->Add(keys[i]);
CutABlock(pib.get(), keys[i]);
VerifyReader(builder.get(), pib.get());
}
void CutABlock(PartitionedIndexBuilder* builder,
const std::string& user_key) {
// Assuming a block is cut, add an entry to the index
std::string key =
std::string(*InternalKey(user_key, 0, ValueType::kTypeValue).rep());
BlockHandle dont_care_block_handle(1, 1);
builder->AddIndexEntry(&key, nullptr, dont_care_block_handle);
}
void CutABlock(PartitionedIndexBuilder* builder, const std::string& user_key,
const std::string& next_user_key) {
// Assuming a block is cut, add an entry to the index
std::string key =
std::string(*InternalKey(user_key, 0, ValueType::kTypeValue).rep());
std::string next_key = std::string(
*InternalKey(next_user_key, 0, ValueType::kTypeValue).rep());
BlockHandle dont_care_block_handle(1, 1);
Slice slice = Slice(next_key.data(), next_key.size());
builder->AddIndexEntry(&key, &slice, dont_care_block_handle);
}
int CountNumOfIndexPartitions(PartitionedIndexBuilder* builder) {
IndexBuilder::IndexBlocks dont_care_ib;
BlockHandle dont_care_bh(10, 10);
Status s;
int cnt = 0;
do {
s = builder->Finish(&dont_care_ib, dont_care_bh);
cnt++;
} while (s.IsIncomplete());
return cnt - 1; // 1 is 2nd level index
}
};
INSTANTIATE_TEST_CASE_P(FormatDef, PartitionedFilterBlockTest,
testing::Values(test::kDefaultFormatVersion));
INSTANTIATE_TEST_CASE_P(FormatLatest, PartitionedFilterBlockTest,
testing::Values(test::kLatestFormatVersion));
TEST_P(PartitionedFilterBlockTest, EmptyBuilder) {
std::unique_ptr<PartitionedIndexBuilder> pib(NewIndexBuilder());
std::unique_ptr<PartitionedFilterBlockBuilder> builder(NewBuilder(pib.get()));
const bool empty = true;
VerifyReader(builder.get(), pib.get(), empty);
}
TEST_P(PartitionedFilterBlockTest, OneBlock) {
uint64_t max_index_size = MaxIndexSize();
for (uint64_t i = 1; i < max_index_size + 1; i++) {
table_options_.metadata_block_size = i;
TestBlockPerAllKeys();
}
}
TEST_P(PartitionedFilterBlockTest, TwoBlocksPerKey) {
uint64_t max_index_size = MaxIndexSize();
for (uint64_t i = 1; i < max_index_size + 1; i++) {
table_options_.metadata_block_size = i;
TestBlockPerTwoKeys();
}
}
// This reproduces the bug that a prefix is the same among multiple consecutive
// blocks but the bug would add it only to the first block.
TEST_P(PartitionedFilterBlockTest, SamePrefixInMultipleBlocks) {
// some small number to cause partition cuts
table_options_.metadata_block_size = 1;
std::unique_ptr<const SliceTransform> prefix_extractor
(rocksdb::NewFixedPrefixTransform(1));
std::unique_ptr<PartitionedIndexBuilder> pib(NewIndexBuilder());
std::unique_ptr<PartitionedFilterBlockBuilder> builder(
NewBuilder(pib.get(), prefix_extractor.get()));
const std::string pkeys[3] = {"p-key10", "p-key20", "p-key30"};
builder->Add(pkeys[0]);
CutABlock(pib.get(), pkeys[0], pkeys[1]);
builder->Add(pkeys[1]);
CutABlock(pib.get(), pkeys[1], pkeys[2]);
builder->Add(pkeys[2]);
CutABlock(pib.get(), pkeys[2]);
std::unique_ptr<PartitionedFilterBlockReader> reader(
NewReader(builder.get(), pib.get()));
for (auto key : pkeys) {
auto ikey = InternalKey(key, 0, ValueType::kTypeValue);
const Slice ikey_slice = Slice(*ikey.rep());
Create a BlockCacheLookupContext to enable fine-grained block cache tracing. (#5421) Summary: BlockCacheLookupContext only contains the caller for now. We will trace block accesses at five places: 1. BlockBasedTable::GetFilter. 2. BlockBasedTable::GetUncompressedDict. 3. BlockBasedTable::MaybeReadAndLoadToCache. (To trace access on data, index, and range deletion block.) 4. BlockBasedTable::Get. (To trace the referenced key and whether the referenced key exists in a fetched data block.) 5. BlockBasedTable::MultiGet. (To trace the referenced key and whether the referenced key exists in a fetched data block.) We create the context at: 1. BlockBasedTable::Get. (kUserGet) 2. BlockBasedTable::MultiGet. (kUserMGet) 3. BlockBasedTable::NewIterator. (either kUserIterator, kCompaction, or external SST ingestion calls this function.) 4. BlockBasedTable::Open. (kPrefetch) 5. Index/Filter::CacheDependencies. (kPrefetch) 6. BlockBasedTable::ApproximateOffsetOf. (kCompaction or kUserApproximateSize). I loaded 1 million key-value pairs into the database and ran the readrandom benchmark with a single thread. I gave the block cache 10 GB to make sure all reads hit the block cache after warmup. The throughput is comparable. Throughput of this PR: 231334 ops/s. Throughput of the master branch: 238428 ops/s. Experiment setup: RocksDB: version 6.2 Date: Mon Jun 10 10:42:51 2019 CPU: 24 * Intel Core Processor (Skylake) CPUCache: 16384 KB Keys: 20 bytes each Values: 100 bytes each (100 bytes after compression) Entries: 1000000 Prefix: 20 bytes Keys per prefix: 0 RawSize: 114.4 MB (estimated) FileSize: 114.4 MB (estimated) Write rate: 0 bytes/second Read rate: 0 ops/second Compression: NoCompression Compression sampling rate: 0 Memtablerep: skip_list Perf Level: 1 Load command: ./db_bench --benchmarks="fillseq" --key_size=20 --prefix_size=20 --keys_per_prefix=0 --value_size=100 --statistics --cache_index_and_filter_blocks --cache_size=10737418240 --disable_auto_compactions=1 --disable_wal=1 --compression_type=none --min_level_to_compress=-1 --compression_ratio=1 --num=1000000 Run command: ./db_bench --benchmarks="readrandom,stats" --use_existing_db --threads=1 --duration=120 --key_size=20 --prefix_size=20 --keys_per_prefix=0 --value_size=100 --statistics --cache_index_and_filter_blocks --cache_size=10737418240 --disable_auto_compactions=1 --disable_wal=1 --compression_type=none --min_level_to_compress=-1 --compression_ratio=1 --num=1000000 --duration=120 TODOs: 1. Create a caller for external SST file ingestion and differentiate the callers for iterator. 2. Integrate tracer to trace block cache accesses. Pull Request resolved: https://github.com/facebook/rocksdb/pull/5421 Differential Revision: D15704258 Pulled By: HaoyuHuang fbshipit-source-id: 4aa8a55f8cb1576ffb367bfa3186a91d8f06d93a
5 years ago
ASSERT_TRUE(reader->PrefixMayMatch(
prefix_extractor->Transform(key), prefix_extractor.get(), kNotValid,
/*no_io=*/false, &ikey_slice, /*get_context=*/nullptr,
/*lookup_context=*/nullptr));
}
// Non-existent keys but with the same prefix
const std::string pnonkeys[4] = {"p-key9", "p-key11", "p-key21", "p-key31"};
for (auto key : pnonkeys) {
auto ikey = InternalKey(key, 0, ValueType::kTypeValue);
const Slice ikey_slice = Slice(*ikey.rep());
ASSERT_TRUE(reader->PrefixMayMatch(
prefix_extractor->Transform(key), prefix_extractor.get(), kNotValid,
/*no_io=*/false, &ikey_slice, /*get_context=*/nullptr,
/*lookup_context=*/nullptr));
}
}
// This reproduces the bug in format_version=3 that the seeking the prefix will
// lead us to the partition before the one that has filter for the prefix.
TEST_P(PartitionedFilterBlockTest, PrefixInWrongPartitionBug) {
// some small number to cause partition cuts
table_options_.metadata_block_size = 1;
std::unique_ptr<const SliceTransform> prefix_extractor(
rocksdb::NewFixedPrefixTransform(2));
std::unique_ptr<PartitionedIndexBuilder> pib(NewIndexBuilder());
std::unique_ptr<PartitionedFilterBlockBuilder> builder(
NewBuilder(pib.get(), prefix_extractor.get()));
// In the bug, searching for prefix "p3" on an index with format version 3,
// will give the key "p3" and the partition of the keys that are <= p3, i.e.,
// p2-keys, where the filter for prefix "p3" does not exist.
const std::string pkeys[] = {"p1-key1", "p2-key2", "p3-key3", "p4-key3",
"p5-key3"};
builder->Add(pkeys[0]);
CutABlock(pib.get(), pkeys[0], pkeys[1]);
builder->Add(pkeys[1]);
CutABlock(pib.get(), pkeys[1], pkeys[2]);
builder->Add(pkeys[2]);
CutABlock(pib.get(), pkeys[2], pkeys[3]);
builder->Add(pkeys[3]);
CutABlock(pib.get(), pkeys[3], pkeys[4]);
builder->Add(pkeys[4]);
CutABlock(pib.get(), pkeys[4]);
std::unique_ptr<PartitionedFilterBlockReader> reader(
NewReader(builder.get(), pib.get()));
for (auto key : pkeys) {
auto prefix = prefix_extractor->Transform(key);
auto ikey = InternalKey(prefix, 0, ValueType::kTypeValue);
const Slice ikey_slice = Slice(*ikey.rep());
ASSERT_TRUE(reader->PrefixMayMatch(
prefix, prefix_extractor.get(), kNotValid,
/*no_io=*/false, &ikey_slice, /*get_context=*/nullptr,
/*lookup_context=*/nullptr));
}
}
TEST_P(PartitionedFilterBlockTest, OneBlockPerKey) {
uint64_t max_index_size = MaxIndexSize();
for (uint64_t i = 1; i < max_index_size + 1; i++) {
table_options_.metadata_block_size = i;
TestBlockPerKey();
}
}
TEST_P(PartitionedFilterBlockTest, PartitionCount) {
int num_keys = sizeof(keys) / sizeof(*keys);
table_options_.metadata_block_size =
std::max(MaxIndexSize(), MaxFilterSize());
int partitions = TestBlockPerKey();
ASSERT_EQ(partitions, 1);
// A low number ensures cutting a block after each key
table_options_.metadata_block_size = 1;
partitions = TestBlockPerKey();
ASSERT_EQ(partitions, num_keys - 1 /* last two keys make one flush */);
}
} // namespace rocksdb
int main(int argc, char** argv) {
::testing::InitGoogleTest(&argc, argv);
return RUN_ALL_TESTS();
}