Revision 0bb555630f5e85a1471843f8dc0dabec297c1c49 authored by Siying Dong on 08 April 2019, 20:24:29 UTC, committed by Facebook Github Bot on 08 April 2019, 20:32:06 UTC
Summary: Create new function NPHash64() and GetSliceNPHash64(), which are currently implemented using murmurhash. Replace the current direct call of murmurhash() to use the new functions if the hash results are not used in on-disk format. This will make it easier to try out or switch to alternative functions in the uses where data format compatibility doesn't need to be considered. This part shouldn't have any performance impact. Also, the sharded cache hash function is changed to the new format, because it falls into this categoery. It doesn't show visible performance impact in db_bench results. CPU showed by perf is increased from about 0.2% to 0.4% in an extreme benchmark setting (4KB blocks, no-compression, everything cached in block cache). We've known that the current hash function used, our own Hash() has serious hash quality problem. It can generate a lots of conflicts with similar input. In this use case, it means extra lock contention for reads from the same file. This slight CPU regression is worthy to me to counter the potential bad performance with hot keys. And hopefully this will get further improved in the future with a better hash function. cache_test's condition is relaxed a little bit to. The new hash is slightly more skewed in this use case, but I manually checked the data and see the hash results are still in a reasonable range. Pull Request resolved: https://github.com/facebook/rocksdb/pull/5155 Differential Revision: D14834821 Pulled By: siying fbshipit-source-id: ec9a2c0a2f8ae4b54d08b13a5c2e9cc97aa80cb5
1 parent de00f28
compaction_picker_universal.h
// 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).
//
// 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
#ifndef ROCKSDB_LITE
#include "db/compaction_picker.h"
namespace rocksdb {
class UniversalCompactionPicker : public CompactionPicker {
public:
UniversalCompactionPicker(const ImmutableCFOptions& ioptions,
const InternalKeyComparator* icmp)
: CompactionPicker(ioptions, icmp) {}
virtual Compaction* PickCompaction(const std::string& cf_name,
const MutableCFOptions& mutable_cf_options,
VersionStorageInfo* vstorage,
LogBuffer* log_buffer) override;
virtual int MaxOutputLevel() const override { return NumberLevels() - 1; }
virtual bool NeedsCompaction(
const VersionStorageInfo* vstorage) const override;
private:
struct SortedRun {
SortedRun(int _level, FileMetaData* _file, uint64_t _size,
uint64_t _compensated_file_size, bool _being_compacted)
: level(_level),
file(_file),
size(_size),
compensated_file_size(_compensated_file_size),
being_compacted(_being_compacted) {
assert(compensated_file_size > 0);
assert(level != 0 || file != nullptr);
}
void Dump(char* out_buf, size_t out_buf_size,
bool print_path = false) const;
// sorted_run_count is added into the string to print
void DumpSizeInfo(char* out_buf, size_t out_buf_size,
size_t sorted_run_count) const;
int level;
// `file` Will be null for level > 0. For level = 0, the sorted run is
// for this file.
FileMetaData* file;
// For level > 0, `size` and `compensated_file_size` are sum of sizes all
// files in the level. `being_compacted` should be the same for all files
// in a non-zero level. Use the value here.
uint64_t size;
uint64_t compensated_file_size;
bool being_compacted;
};
// Pick Universal compaction to limit read amplification
Compaction* PickCompactionToReduceSortedRuns(
const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
VersionStorageInfo* vstorage, double score, unsigned int ratio,
unsigned int num_files, const std::vector<SortedRun>& sorted_runs,
LogBuffer* log_buffer);
// Pick Universal compaction to limit space amplification.
Compaction* PickCompactionToReduceSizeAmp(
const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
VersionStorageInfo* vstorage, double score,
const std::vector<SortedRun>& sorted_runs, LogBuffer* log_buffer);
Compaction* PickDeleteTriggeredCompaction(
const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
VersionStorageInfo* vstorage, double score,
const std::vector<SortedRun>& sorted_runs, LogBuffer* log_buffer);
// Used in universal compaction when the enabled_trivial_move
// option is set. Checks whether there are any overlapping files
// in the input. Returns true if the input files are non
// overlapping.
bool IsInputFilesNonOverlapping(Compaction* c);
static std::vector<SortedRun> CalculateSortedRuns(
const VersionStorageInfo& vstorage, const ImmutableCFOptions& ioptions,
const MutableCFOptions& mutable_cf_options);
// Pick a path ID to place a newly generated file, with its estimated file
// size.
static uint32_t GetPathId(const ImmutableCFOptions& ioptions,
const MutableCFOptions& mutable_cf_options,
uint64_t file_size);
};
} // namespace rocksdb
#endif // !ROCKSDB_LITE
![swh spinner](/static/img/swh-spinner.gif)
Computing file changes ...