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_table_reader.h

232 lines
9.0 KiB

// Copyright (c) 2013, Facebook, Inc. All rights reserved.
// This source code is licensed under the BSD-style license found in the
// LICENSE file in the root directory of this source tree. An additional grant
// of patent rights can be found in the PATENTS file in the same directory.
//
// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file. See the AUTHORS file for names of contributors.
#pragma once
#include <stdint.h>
#include <memory>
#include <utility>
#include <string>
#include "rocksdb/options.h"
#include "rocksdb/statistics.h"
#include "rocksdb/status.h"
#include "rocksdb/table.h"
#include "table/table_reader.h"
#include "table/table_properties_internal.h"
#include "util/coding.h"
#include "util/file_reader_writer.h"
namespace rocksdb {
class Block;
class BlockIter;
class BlockHandle;
class Cache;
class FilterBlockReader;
Implement full filter for block based table. Summary: 1. Make filter_block.h a base class. Derive block_based_filter_block and full_filter_block. The previous one is the traditional filter block. The full_filter_block is newly added. It would generate a filter block that contain all the keys in SST file. 2. When querying a key, table would first check if full_filter is available. If not, it would go to the exact data block and check using block_based filter. 3. User could choose to use full_filter or tradional(block_based_filter). They would be stored in SST file with different meta index name. "filter.filter_policy" or "full_filter.filter_policy". Then, Table reader is able to know the fllter block type. 4. Some optimizations have been done for full_filter_block, thus it requires a different interface compared to the original one in filter_policy.h. 5. Actual implementation of filter bits coding/decoding is placed in util/bloom_impl.cc Benchmark: base commit 1d23b5c470844c1208301311f0889eca750431c0 Command: db_bench --db=/dev/shm/rocksdb --num_levels=6 --key_size=20 --prefix_size=20 --keys_per_prefix=0 --value_size=100 --write_buffer_size=134217728 --max_write_buffer_number=2 --target_file_size_base=33554432 --max_bytes_for_level_base=1073741824 --verify_checksum=false --max_background_compactions=4 --use_plain_table=0 --memtablerep=prefix_hash --open_files=-1 --mmap_read=1 --mmap_write=0 --bloom_bits=10 --bloom_locality=1 --memtable_bloom_bits=500000 --compression_type=lz4 --num=393216000 --use_hash_search=1 --block_size=1024 --block_restart_interval=16 --use_existing_db=1 --threads=1 --benchmarks=readrandom —disable_auto_compactions=1 Read QPS increase for about 30% from 2230002 to 2991411. Test Plan: make all check valgrind db_test db_stress --use_block_based_filter = 0 ./auto_sanity_test.sh Reviewers: igor, yhchiang, ljin, sdong Reviewed By: sdong Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D20979
10 years ago
class BlockBasedFilterBlockReader;
class FullFilterBlockReader;
class Footer;
class InternalKeyComparator;
class Iterator;
class RandomAccessFile;
class TableCache;
class TableReader;
class WritableFile;
struct BlockBasedTableOptions;
struct EnvOptions;
struct ReadOptions;
class GetContext;
using std::unique_ptr;
// A Table is a sorted map from strings to strings. Tables are
// immutable and persistent. A Table may be safely accessed from
// multiple threads without external synchronization.
class BlockBasedTable : public TableReader {
public:
static const std::string kFilterBlockPrefix;
Implement full filter for block based table. Summary: 1. Make filter_block.h a base class. Derive block_based_filter_block and full_filter_block. The previous one is the traditional filter block. The full_filter_block is newly added. It would generate a filter block that contain all the keys in SST file. 2. When querying a key, table would first check if full_filter is available. If not, it would go to the exact data block and check using block_based filter. 3. User could choose to use full_filter or tradional(block_based_filter). They would be stored in SST file with different meta index name. "filter.filter_policy" or "full_filter.filter_policy". Then, Table reader is able to know the fllter block type. 4. Some optimizations have been done for full_filter_block, thus it requires a different interface compared to the original one in filter_policy.h. 5. Actual implementation of filter bits coding/decoding is placed in util/bloom_impl.cc Benchmark: base commit 1d23b5c470844c1208301311f0889eca750431c0 Command: db_bench --db=/dev/shm/rocksdb --num_levels=6 --key_size=20 --prefix_size=20 --keys_per_prefix=0 --value_size=100 --write_buffer_size=134217728 --max_write_buffer_number=2 --target_file_size_base=33554432 --max_bytes_for_level_base=1073741824 --verify_checksum=false --max_background_compactions=4 --use_plain_table=0 --memtablerep=prefix_hash --open_files=-1 --mmap_read=1 --mmap_write=0 --bloom_bits=10 --bloom_locality=1 --memtable_bloom_bits=500000 --compression_type=lz4 --num=393216000 --use_hash_search=1 --block_size=1024 --block_restart_interval=16 --use_existing_db=1 --threads=1 --benchmarks=readrandom —disable_auto_compactions=1 Read QPS increase for about 30% from 2230002 to 2991411. Test Plan: make all check valgrind db_test db_stress --use_block_based_filter = 0 ./auto_sanity_test.sh Reviewers: igor, yhchiang, ljin, sdong Reviewed By: sdong Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D20979
10 years ago
static const std::string kFullFilterBlockPrefix;
// Attempt to open the table that is stored in bytes [0..file_size)
// of "file", and read the metadata entries necessary to allow
// retrieving data from the table.
//
// If successful, returns ok and sets "*table_reader" to the newly opened
// table. The client should delete "*table_reader" when no longer needed.
// If there was an error while initializing the table, sets "*table_reader"
// to nullptr and returns a non-ok status.
//
// *file must remain live while this Table is in use.
Disable pre-fetching of index and filter blocks for sst_dump_tool. Summary: BlockBasedTable pre-fetches the filter and index blocks on Open call. This is an optimistic optimization targeted for runtime scenario. The optimization is unnecessary for sst_dump_tool - Added a provision to disable pre-fetching of index and filter blocks in BlockBasedTable - Disabled pre-fetching for the sst_dump tool Stack for reference : #01 0x00000000005ed944 in snappy::InternalUncompress<snappy::SnappyArrayWriter> () from /home/engshare/third-party2/snappy/1.0.3/src/snappy-1.0.3/snappy.cc:148 #02 0x00000000005edeee in snappy::RawUncompress () from /home/engshare/third-party2/snappy/1.0.3/src/snappy-1.0.3/snappy.cc:947 #03 0x00000000004e0b4d in rocksdb::UncompressBlockContents () from /data/users/paultuckfield/rocksdb/./util/compression.h:69 #04 0x00000000004e145c in rocksdb::ReadBlockContents () from /data/users/paultuckfield/rocksdb/table/format.cc:334 #05 0x00000000004ca424 in rocksdb::(anonymous namespace)::ReadBlockFromFile () from /data/users/paultuckfield/rocksdb/table/block_based_table_reader.cc:70 #06 0x00000000004cccad in rocksdb::BlockBasedTable::CreateIndexReader () from /data/users/paultuckfield/rocksdb/table/block_based_table_reader.cc:173 #07 0x00000000004d17e5 in rocksdb::BlockBasedTable::Open () from /data/users/paultuckfield/rocksdb/table/block_based_table_reader.cc:553 #08 0x00000000004c8184 in rocksdb::BlockBasedTableFactory::NewTableReader () from /data/users/paultuckfield/rocksdb/table/block_based_table_factory.cc:51 #09 0x0000000000598463 in rocksdb::SstFileReader::NewTableReader () from /data/users/paultuckfield/rocksdb/util/sst_dump_tool.cc:69 #10 0x00000000005986c2 in rocksdb::SstFileReader::SstFileReader () from /data/users/paultuckfield/rocksdb/util/sst_dump_tool.cc:26 #11 0x0000000000599047 in rocksdb::SSTDumpTool::Run () from /data/users/paultuckfield/rocksdb/util/sst_dump_tool.cc:332 #12 0x0000000000409b06 in main () from /data/users/paultuckfield/rocksdb/tools/sst_dump.cc:12 Test Plan: - Added a unit test to trigger the code. - Also did some manual verification. - Passed all unit tests task #6296048 Reviewers: igor, rven, sdong Reviewed By: sdong Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D34041
10 years ago
// *prefetch_blocks can be used to disable prefetching of index and filter
// blocks at statup
static Status Open(const ImmutableCFOptions& ioptions,
const EnvOptions& env_options,
const BlockBasedTableOptions& table_options,
const InternalKeyComparator& internal_key_comparator,
unique_ptr<RandomAccessFileReader>&& file,
uint64_t file_size, unique_ptr<TableReader>* table_reader,
Disable pre-fetching of index and filter blocks for sst_dump_tool. Summary: BlockBasedTable pre-fetches the filter and index blocks on Open call. This is an optimistic optimization targeted for runtime scenario. The optimization is unnecessary for sst_dump_tool - Added a provision to disable pre-fetching of index and filter blocks in BlockBasedTable - Disabled pre-fetching for the sst_dump tool Stack for reference : #01 0x00000000005ed944 in snappy::InternalUncompress<snappy::SnappyArrayWriter> () from /home/engshare/third-party2/snappy/1.0.3/src/snappy-1.0.3/snappy.cc:148 #02 0x00000000005edeee in snappy::RawUncompress () from /home/engshare/third-party2/snappy/1.0.3/src/snappy-1.0.3/snappy.cc:947 #03 0x00000000004e0b4d in rocksdb::UncompressBlockContents () from /data/users/paultuckfield/rocksdb/./util/compression.h:69 #04 0x00000000004e145c in rocksdb::ReadBlockContents () from /data/users/paultuckfield/rocksdb/table/format.cc:334 #05 0x00000000004ca424 in rocksdb::(anonymous namespace)::ReadBlockFromFile () from /data/users/paultuckfield/rocksdb/table/block_based_table_reader.cc:70 #06 0x00000000004cccad in rocksdb::BlockBasedTable::CreateIndexReader () from /data/users/paultuckfield/rocksdb/table/block_based_table_reader.cc:173 #07 0x00000000004d17e5 in rocksdb::BlockBasedTable::Open () from /data/users/paultuckfield/rocksdb/table/block_based_table_reader.cc:553 #08 0x00000000004c8184 in rocksdb::BlockBasedTableFactory::NewTableReader () from /data/users/paultuckfield/rocksdb/table/block_based_table_factory.cc:51 #09 0x0000000000598463 in rocksdb::SstFileReader::NewTableReader () from /data/users/paultuckfield/rocksdb/util/sst_dump_tool.cc:69 #10 0x00000000005986c2 in rocksdb::SstFileReader::SstFileReader () from /data/users/paultuckfield/rocksdb/util/sst_dump_tool.cc:26 #11 0x0000000000599047 in rocksdb::SSTDumpTool::Run () from /data/users/paultuckfield/rocksdb/util/sst_dump_tool.cc:332 #12 0x0000000000409b06 in main () from /data/users/paultuckfield/rocksdb/tools/sst_dump.cc:12 Test Plan: - Added a unit test to trigger the code. - Also did some manual verification. - Passed all unit tests task #6296048 Reviewers: igor, rven, sdong Reviewed By: sdong Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D34041
10 years ago
bool prefetch_index_and_filter = true);
bool PrefixMayMatch(const Slice& internal_key);
// Returns a new iterator over the table contents.
// The result of NewIterator() is initially invalid (caller must
// call one of the Seek methods on the iterator before using it).
Iterator* NewIterator(const ReadOptions&, Arena* arena = nullptr) override;
Status Get(const ReadOptions& readOptions, const Slice& key,
GetContext* get_context) override;
// Pre-fetch the disk blocks that correspond to the key range specified by
// (kbegin, kend). The call will return return error status in the event of
// IO or iteration error.
Status Prefetch(const Slice* begin, const Slice* end) override;
// Given a key, return an approximate byte offset in the file where
// the data for that key begins (or would begin if the key were
// present in the file). The returned value is in terms of file
// bytes, and so includes effects like compression of the underlying data.
// E.g., the approximate offset of the last key in the table will
// be close to the file length.
uint64_t ApproximateOffsetOf(const Slice& key) override;
// Returns true if the block for the specified key is in cache.
// REQUIRES: key is in this table && block cache enabled
bool TEST_KeyInCache(const ReadOptions& options, const Slice& key);
// Set up the table for Compaction. Might change some parameters with
// posix_fadvise
void SetupForCompaction() override;
std::shared_ptr<const TableProperties> GetTableProperties() const override;
size_t ApproximateMemoryUsage() const override;
// convert SST file to a human readable form
Status DumpTable(WritableFile* out_file) override;
~BlockBasedTable();
bool TEST_filter_block_preloaded() const;
bool TEST_index_reader_preloaded() const;
// Implementation of IndexReader will be exposed to internal cc file only.
class IndexReader;
private:
template <class TValue>
struct CachableEntry;
struct Rep;
Rep* rep_;
bool compaction_optimized_;
11 years ago
class BlockEntryIteratorState;
// input_iter: if it is not null, update this one and return it as Iterator
static Iterator* NewDataBlockIterator(Rep* rep, const ReadOptions& ro,
const Slice& index_value,
BlockIter* input_iter = nullptr);
// For the following two functions:
// if `no_io == true`, we will not try to read filter/index from sst file
// were they not present in cache yet.
CachableEntry<FilterBlockReader> GetFilter(bool no_io = false) const;
// Get the iterator from the index reader.
// If input_iter is not set, return new Iterator
// If input_iter is set, update it and return it as Iterator
//
// Note: ErrorIterator with Status::Incomplete shall be returned if all the
// following conditions are met:
// 1. We enabled table_options.cache_index_and_filter_blocks.
// 2. index is not present in block cache.
// 3. We disallowed any io to be performed, that is, read_options ==
// kBlockCacheTier
Iterator* NewIndexIterator(const ReadOptions& read_options,
BlockIter* input_iter = nullptr);
// Read block cache from block caches (if set): block_cache and
// block_cache_compressed.
// On success, Status::OK with be returned and @block will be populated with
// pointer to the block as well as its block handle.
static Status GetDataBlockFromCache(
const Slice& block_cache_key, const Slice& compressed_block_cache_key,
Cache* block_cache, Cache* block_cache_compressed, Statistics* statistics,
const ReadOptions& read_options,
BlockBasedTable::CachableEntry<Block>* block, uint32_t format_version);
// Put a raw block (maybe compressed) to the corresponding block caches.
// This method will perform decompression against raw_block if needed and then
// populate the block caches.
// On success, Status::OK will be returned; also @block will be populated with
// uncompressed block and its cache handle.
//
// REQUIRES: raw_block is heap-allocated. PutDataBlockToCache() will be
// responsible for releasing its memory if error occurs.
static Status PutDataBlockToCache(
const Slice& block_cache_key, const Slice& compressed_block_cache_key,
Cache* block_cache, Cache* block_cache_compressed,
const ReadOptions& read_options, Statistics* statistics,
CachableEntry<Block>* block, Block* raw_block, uint32_t format_version);
// Calls (*handle_result)(arg, ...) repeatedly, starting with the entry found
// after a call to Seek(key), until handle_result returns false.
// May not make such a call if filter policy says that key is not present.
friend class TableCache;
friend class BlockBasedTableBuilder;
void ReadMeta(const Footer& footer);
// Create a index reader based on the index type stored in the table.
// Optionally, user can pass a preloaded meta_index_iter for the index that
// need to access extra meta blocks for index construction. This parameter
// helps avoid re-reading meta index block if caller already created one.
Status CreateIndexReader(IndexReader** index_reader,
Iterator* preloaded_meta_index_iter = nullptr);
bool FullFilterKeyMayMatch(FilterBlockReader* filter,
const Slice& user_key) const;
// Read the meta block from sst.
static Status ReadMetaBlock(
Rep* rep,
std::unique_ptr<Block>* meta_block,
std::unique_ptr<Iterator>* iter);
// Create the filter from the filter block.
static FilterBlockReader* ReadFilter(
Rep* rep,
Iterator* meta_index_iter,
size_t* filter_size = nullptr);
static void SetupCacheKeyPrefix(Rep* rep);
explicit BlockBasedTable(Rep* rep)
: rep_(rep), compaction_optimized_(false) {}
// Generate a cache key prefix from the file
static void GenerateCachePrefix(Cache* cc,
RandomAccessFile* file, char* buffer, size_t* size);
static void GenerateCachePrefix(Cache* cc,
WritableFile* file, char* buffer, size_t* size);
// The longest prefix of the cache key used to identify blocks.
// For Posix files the unique ID is three varints.
static const size_t kMaxCacheKeyPrefixSize = kMaxVarint64Length*3+1;
// Helper functions for DumpTable()
Status DumpIndexBlock(WritableFile* out_file);
Status DumpDataBlocks(WritableFile* out_file);
// No copying allowed
explicit BlockBasedTable(const TableReader&) = delete;
void operator=(const TableReader&) = delete;
};
} // namespace rocksdb