rocksdb/cache/cache_key.h

135 lines
5.3 KiB
C
Raw Normal View History

New stable, fixed-length cache keys (#9126) Summary: This change standardizes on a new 16-byte cache key format for block cache (incl compressed and secondary) and persistent cache (but not table cache and row cache). The goal is a really fast cache key with practically ideal stability and uniqueness properties without external dependencies (e.g. from FileSystem). A fixed key size of 16 bytes should enable future optimizations to the concurrent hash table for block cache, which is a heavy CPU user / bottleneck, but there appears to be measurable performance improvement even with no changes to LRUCache. This change replaces a lot of disjointed and ugly code handling cache keys with calls to a simple, clean new internal API (cache_key.h). (Preserving the old cache key logic under an option would be very ugly and likely negate the performance gain of the new approach. Complete replacement carries some inherent risk, but I think that's acceptable with sufficient analysis and testing.) The scheme for encoding new cache keys is complicated but explained in cache_key.cc. Also: EndianSwapValue is moved to math.h to be next to other bit operations. (Explains some new include "math.h".) ReverseBits operation added and unit tests added to hash_test for both. Fixes https://github.com/facebook/rocksdb/issues/7405 (presuming a root cause) Pull Request resolved: https://github.com/facebook/rocksdb/pull/9126 Test Plan: ### Basic correctness Several tests needed updates to work with the new functionality, mostly because we are no longer relying on filesystem for stable cache keys so table builders & readers need more context info to agree on cache keys. This functionality is so core, a huge number of existing tests exercise the cache key functionality. ### Performance Create db with `TEST_TMPDIR=/dev/shm ./db_bench -bloom_bits=10 -benchmarks=fillrandom -num=3000000 -partition_index_and_filters` And test performance with `TEST_TMPDIR=/dev/shm ./db_bench -readonly -use_existing_db -bloom_bits=10 -benchmarks=readrandom -num=3000000 -duration=30 -cache_index_and_filter_blocks -cache_size=250000 -threads=4` using DEBUG_LEVEL=0 and simultaneous before & after runs. Before ops/sec, avg over 100 runs: 121924 After ops/sec, avg over 100 runs: 125385 (+2.8%) ### Collision probability I have built a tool, ./cache_bench -stress_cache_key to broadly simulate host-wide cache activity over many months, by making some pessimistic simplifying assumptions: * Every generated file has a cache entry for every byte offset in the file (contiguous range of cache keys) * All of every file is cached for its entire lifetime We use a simple table with skewed address assignment and replacement on address collision to simulate files coming & going, with quite a variance (super-Poisson) in ages. Some output with `./cache_bench -stress_cache_key -sck_keep_bits=40`: ``` Total cache or DBs size: 32TiB Writing 925.926 MiB/s or 76.2939TiB/day Multiply by 9.22337e+18 to correct for simulation losses (but still assume whole file cached) ``` These come from default settings of 2.5M files per day of 32 MB each, and `-sck_keep_bits=40` means that to represent a single file, we are only keeping 40 bits of the 128-bit cache key. With file size of 2\*\*25 contiguous keys (pessimistic), our simulation is about 2\*\*(128-40-25) or about 9 billion billion times more prone to collision than reality. More default assumptions, relatively pessimistic: * 100 DBs in same process (doesn't matter much) * Re-open DB in same process (new session ID related to old session ID) on average every 100 files generated * Restart process (all new session IDs unrelated to old) 24 times per day After enough data, we get a result at the end: ``` (keep 40 bits) 17 collisions after 2 x 90 days, est 10.5882 days between (9.76592e+19 corrected) ``` If we believe the (pessimistic) simulation and the mathematical generalization, we would need to run a billion machines all for 97 billion days to expect a cache key collision. To help verify that our generalization ("corrected") is robust, we can make our simulation more precise with `-sck_keep_bits=41` and `42`, which takes more running time to get enough data: ``` (keep 41 bits) 16 collisions after 4 x 90 days, est 22.5 days between (1.03763e+20 corrected) (keep 42 bits) 19 collisions after 10 x 90 days, est 47.3684 days between (1.09224e+20 corrected) ``` The generalized prediction still holds. With the `-sck_randomize` option, we can see that we are beating "random" cache keys (except offsets still non-randomized) by a modest amount (roughly 20x less collision prone than random), which should make us reasonably comfortable even in "degenerate" cases: ``` 197 collisions after 1 x 90 days, est 0.456853 days between (4.21372e+18 corrected) ``` I've run other tests to validate other conditions behave as expected, never behaving "worse than random" unless we start chopping off structured data. Reviewed By: zhichao-cao Differential Revision: D33171746 Pulled By: pdillinger fbshipit-source-id: f16a57e369ed37be5e7e33525ace848d0537c88f
2021-12-17 01:13:55 +00:00
// Copyright (c) Facebook, Inc. and its affiliates. 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).
#pragma once
#include <cstdint>
#include "rocksdb/rocksdb_namespace.h"
#include "rocksdb/slice.h"
namespace ROCKSDB_NAMESPACE {
class Cache;
// A standard holder for fixed-size block cache keys (and for related caches).
// They are created through one of these, each using its own range of values:
// * CacheKey::CreateUniqueForCacheLifetime
// * CacheKey::CreateUniqueForProcessLifetime
// * Default ctor ("empty" cache key)
// * OffsetableCacheKey->WithOffset
//
// The first two use atomic counters to guarantee uniqueness over the given
// lifetime and the last uses a form of universally unique identifier for
// uniqueness with very high probabilty (and guaranteed for files generated
// during a single process lifetime).
//
// CacheKeys are currently used by calling AsSlice() to pass as a key to
// Cache. For performance, the keys are endianness-dependent (though otherwise
// portable). (Persistable cache entries are not intended to cross platforms.)
class CacheKey {
public:
// For convenience, constructs an "empty" cache key that is never returned
// by other means.
inline CacheKey() : session_etc64_(), offset_etc64_() {}
inline bool IsEmpty() const {
return (session_etc64_ == 0) & (offset_etc64_ == 0);
}
// Use this cache key as a Slice (byte order is endianness-dependent)
inline Slice AsSlice() const {
static_assert(sizeof(*this) == 16, "Standardized on 16-byte cache key");
assert(!IsEmpty());
return Slice(reinterpret_cast<const char *>(this), sizeof(*this));
}
// Create a CacheKey that is unique among others associated with this Cache
// instance. Depends on Cache::NewId. This is useful for block cache
// "reservations".
static CacheKey CreateUniqueForCacheLifetime(Cache *cache);
// Create a CacheKey that is unique among others for the lifetime of this
// process. This is useful for saving in a static data member so that
// different DB instances can agree on a cache key for shared entities,
// such as for CacheEntryStatsCollector.
static CacheKey CreateUniqueForProcessLifetime();
protected:
friend class OffsetableCacheKey;
CacheKey(uint64_t session_etc64, uint64_t offset_etc64)
: session_etc64_(session_etc64), offset_etc64_(offset_etc64) {}
uint64_t session_etc64_;
uint64_t offset_etc64_;
};
constexpr uint8_t kCacheKeySize = static_cast<uint8_t>(sizeof(CacheKey));
New stable, fixed-length cache keys (#9126) Summary: This change standardizes on a new 16-byte cache key format for block cache (incl compressed and secondary) and persistent cache (but not table cache and row cache). The goal is a really fast cache key with practically ideal stability and uniqueness properties without external dependencies (e.g. from FileSystem). A fixed key size of 16 bytes should enable future optimizations to the concurrent hash table for block cache, which is a heavy CPU user / bottleneck, but there appears to be measurable performance improvement even with no changes to LRUCache. This change replaces a lot of disjointed and ugly code handling cache keys with calls to a simple, clean new internal API (cache_key.h). (Preserving the old cache key logic under an option would be very ugly and likely negate the performance gain of the new approach. Complete replacement carries some inherent risk, but I think that's acceptable with sufficient analysis and testing.) The scheme for encoding new cache keys is complicated but explained in cache_key.cc. Also: EndianSwapValue is moved to math.h to be next to other bit operations. (Explains some new include "math.h".) ReverseBits operation added and unit tests added to hash_test for both. Fixes https://github.com/facebook/rocksdb/issues/7405 (presuming a root cause) Pull Request resolved: https://github.com/facebook/rocksdb/pull/9126 Test Plan: ### Basic correctness Several tests needed updates to work with the new functionality, mostly because we are no longer relying on filesystem for stable cache keys so table builders & readers need more context info to agree on cache keys. This functionality is so core, a huge number of existing tests exercise the cache key functionality. ### Performance Create db with `TEST_TMPDIR=/dev/shm ./db_bench -bloom_bits=10 -benchmarks=fillrandom -num=3000000 -partition_index_and_filters` And test performance with `TEST_TMPDIR=/dev/shm ./db_bench -readonly -use_existing_db -bloom_bits=10 -benchmarks=readrandom -num=3000000 -duration=30 -cache_index_and_filter_blocks -cache_size=250000 -threads=4` using DEBUG_LEVEL=0 and simultaneous before & after runs. Before ops/sec, avg over 100 runs: 121924 After ops/sec, avg over 100 runs: 125385 (+2.8%) ### Collision probability I have built a tool, ./cache_bench -stress_cache_key to broadly simulate host-wide cache activity over many months, by making some pessimistic simplifying assumptions: * Every generated file has a cache entry for every byte offset in the file (contiguous range of cache keys) * All of every file is cached for its entire lifetime We use a simple table with skewed address assignment and replacement on address collision to simulate files coming & going, with quite a variance (super-Poisson) in ages. Some output with `./cache_bench -stress_cache_key -sck_keep_bits=40`: ``` Total cache or DBs size: 32TiB Writing 925.926 MiB/s or 76.2939TiB/day Multiply by 9.22337e+18 to correct for simulation losses (but still assume whole file cached) ``` These come from default settings of 2.5M files per day of 32 MB each, and `-sck_keep_bits=40` means that to represent a single file, we are only keeping 40 bits of the 128-bit cache key. With file size of 2\*\*25 contiguous keys (pessimistic), our simulation is about 2\*\*(128-40-25) or about 9 billion billion times more prone to collision than reality. More default assumptions, relatively pessimistic: * 100 DBs in same process (doesn't matter much) * Re-open DB in same process (new session ID related to old session ID) on average every 100 files generated * Restart process (all new session IDs unrelated to old) 24 times per day After enough data, we get a result at the end: ``` (keep 40 bits) 17 collisions after 2 x 90 days, est 10.5882 days between (9.76592e+19 corrected) ``` If we believe the (pessimistic) simulation and the mathematical generalization, we would need to run a billion machines all for 97 billion days to expect a cache key collision. To help verify that our generalization ("corrected") is robust, we can make our simulation more precise with `-sck_keep_bits=41` and `42`, which takes more running time to get enough data: ``` (keep 41 bits) 16 collisions after 4 x 90 days, est 22.5 days between (1.03763e+20 corrected) (keep 42 bits) 19 collisions after 10 x 90 days, est 47.3684 days between (1.09224e+20 corrected) ``` The generalized prediction still holds. With the `-sck_randomize` option, we can see that we are beating "random" cache keys (except offsets still non-randomized) by a modest amount (roughly 20x less collision prone than random), which should make us reasonably comfortable even in "degenerate" cases: ``` 197 collisions after 1 x 90 days, est 0.456853 days between (4.21372e+18 corrected) ``` I've run other tests to validate other conditions behave as expected, never behaving "worse than random" unless we start chopping off structured data. Reviewed By: zhichao-cao Differential Revision: D33171746 Pulled By: pdillinger fbshipit-source-id: f16a57e369ed37be5e7e33525ace848d0537c88f
2021-12-17 01:13:55 +00:00
// A file-specific generator of cache keys, sometimes referred to as the
// "base" cache key for a file because all the cache keys for various offsets
// within the file are computed using simple arithmetic. The basis for the
// general approach is dicussed here: https://github.com/pdillinger/unique_id
// Heavily related to GetUniqueIdFromTableProperties.
//
// If the db_id, db_session_id, and file_number come from the file's table
// properties, then the keys will be stable across DB::Open/Close, backup/
// restore, import/export, etc.
//
// This class "is a" CacheKey only privately so that it is not misused as
// a ready-to-use CacheKey.
class OffsetableCacheKey : private CacheKey {
public:
// For convenience, constructs an "empty" cache key that should not be used.
inline OffsetableCacheKey() : CacheKey() {}
// Constructs an OffsetableCacheKey with the given information about a file.
// max_offset is based on file size (see WithOffset) and is required here to
// choose an appropriate (sub-)encoding. This constructor never generates an
// "empty" base key.
OffsetableCacheKey(const std::string &db_id, const std::string &db_session_id,
uint64_t file_number, uint64_t max_offset);
inline bool IsEmpty() const {
bool result = session_etc64_ == 0;
assert(!(offset_etc64_ > 0 && result));
return result;
}
// Construct a CacheKey for an offset within a file, which must be
// <= max_offset provided in constructor. An offset is not necessarily a
// byte offset if a smaller unique identifier of keyable offsets is used.
//
// This class was designed to make this hot code extremely fast.
inline CacheKey WithOffset(uint64_t offset) const {
assert(!IsEmpty());
assert(offset <= max_offset_);
return CacheKey(session_etc64_, offset_etc64_ ^ offset);
}
// The "common prefix" is a shared prefix for all the returned CacheKeys,
// that also happens to usually be the same among many files in the same DB,
// so is efficient and highly accurate (not perfectly) for DB-specific cache
// dump selection (but not file-specific).
static constexpr size_t kCommonPrefixSize = 8;
inline Slice CommonPrefixSlice() const {
static_assert(sizeof(session_etc64_) == kCommonPrefixSize,
"8 byte common prefix expected");
assert(!IsEmpty());
assert(&this->session_etc64_ == static_cast<const void *>(this));
return Slice(reinterpret_cast<const char *>(this), kCommonPrefixSize);
}
// For any max_offset <= this value, the same encoding scheme is guaranteed.
static constexpr uint64_t kMaxOffsetStandardEncoding = 0xffffffffffU;
private:
#ifndef NDEBUG
uint64_t max_offset_ = 0;
#endif
};
} // namespace ROCKSDB_NAMESPACE