rocksdb/db/external_sst_file_ingestion_job.h
Peter Dillinger 0050a73a4f 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
2021-12-16 17:15:13 -08:00

200 lines
7.5 KiB
C++

// 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).
#pragma once
#include <string>
#include <unordered_set>
#include <vector>
#include "db/column_family.h"
#include "db/internal_stats.h"
#include "db/snapshot_impl.h"
#include "env/file_system_tracer.h"
#include "logging/event_logger.h"
#include "options/db_options.h"
#include "rocksdb/db.h"
#include "rocksdb/file_system.h"
#include "rocksdb/sst_file_writer.h"
#include "util/autovector.h"
namespace ROCKSDB_NAMESPACE {
class Directories;
class SystemClock;
struct IngestedFileInfo {
// External file path
std::string external_file_path;
// Smallest internal key in external file
InternalKey smallest_internal_key;
// Largest internal key in external file
InternalKey largest_internal_key;
// Sequence number for keys in external file
SequenceNumber original_seqno;
// Offset of the global sequence number field in the file, will
// be zero if version is 1 (global seqno is not supported)
size_t global_seqno_offset;
// External file size
uint64_t file_size;
// total number of keys in external file
uint64_t num_entries;
// total number of range deletions in external file
uint64_t num_range_deletions;
// Id of column family this file shoule be ingested into
uint32_t cf_id;
// TableProperties read from external file
TableProperties table_properties;
// Version of external file
int version;
// FileDescriptor for the file inside the DB
FileDescriptor fd;
// file path that we picked for file inside the DB
std::string internal_file_path;
// Global sequence number that we picked for the file inside the DB
SequenceNumber assigned_seqno = 0;
// Level inside the DB we picked for the external file.
int picked_level = 0;
// Whether to copy or link the external sst file. copy_file will be set to
// false if ingestion_options.move_files is true and underlying FS
// supports link operation. Need to provide a default value to make the
// undefined-behavior sanity check of llvm happy. Since
// ingestion_options.move_files is false by default, thus copy_file is true
// by default.
bool copy_file = true;
// The checksum of ingested file
std::string file_checksum;
// The name of checksum function that generate the checksum
std::string file_checksum_func_name;
// The temperature of the file to be ingested
Temperature file_temperature = Temperature::kUnknown;
};
class ExternalSstFileIngestionJob {
public:
ExternalSstFileIngestionJob(
VersionSet* versions, ColumnFamilyData* cfd,
const ImmutableDBOptions& db_options, const EnvOptions& env_options,
SnapshotList* db_snapshots,
const IngestExternalFileOptions& ingestion_options,
Directories* directories, EventLogger* event_logger,
const std::shared_ptr<IOTracer>& io_tracer)
: clock_(db_options.clock),
fs_(db_options.fs, io_tracer),
versions_(versions),
cfd_(cfd),
db_options_(db_options),
env_options_(env_options),
db_snapshots_(db_snapshots),
ingestion_options_(ingestion_options),
directories_(directories),
event_logger_(event_logger),
job_start_time_(clock_->NowMicros()),
consumed_seqno_count_(0),
io_tracer_(io_tracer) {
assert(directories != nullptr);
}
// Prepare the job by copying external files into the DB.
Status Prepare(const std::vector<std::string>& external_files_paths,
const std::vector<std::string>& files_checksums,
const std::vector<std::string>& files_checksum_func_names,
const Temperature& file_temperature, uint64_t next_file_number,
SuperVersion* sv);
// Check if we need to flush the memtable before running the ingestion job
// This will be true if the files we are ingesting are overlapping with any
// key range in the memtable.
//
// @param super_version A referenced SuperVersion that will be held for the
// duration of this function.
//
// Thread-safe
Status NeedsFlush(bool* flush_needed, SuperVersion* super_version);
// Will execute the ingestion job and prepare edit() to be applied.
// REQUIRES: Mutex held
Status Run();
// Update column family stats.
// REQUIRES: Mutex held
void UpdateStats();
// Cleanup after successful/failed job
void Cleanup(const Status& status);
VersionEdit* edit() { return &edit_; }
const autovector<IngestedFileInfo>& files_to_ingest() const {
return files_to_ingest_;
}
// How many sequence numbers did we consume as part of the ingest job?
int ConsumedSequenceNumbersCount() const { return consumed_seqno_count_; }
private:
// Open the external file and populate `file_to_ingest` with all the
// external information we need to ingest this file.
Status GetIngestedFileInfo(const std::string& external_file,
uint64_t new_file_number,
IngestedFileInfo* file_to_ingest,
SuperVersion* sv);
// Assign `file_to_ingest` the appropriate sequence number and the lowest
// possible level that it can be ingested to according to compaction_style.
// REQUIRES: Mutex held
Status AssignLevelAndSeqnoForIngestedFile(SuperVersion* sv,
bool force_global_seqno,
CompactionStyle compaction_style,
SequenceNumber last_seqno,
IngestedFileInfo* file_to_ingest,
SequenceNumber* assigned_seqno);
// File that we want to ingest behind always goes to the lowest level;
// we just check that it fits in the level, that DB allows ingest_behind,
// and that we don't have 0 seqnums at the upper levels.
// REQUIRES: Mutex held
Status CheckLevelForIngestedBehindFile(IngestedFileInfo* file_to_ingest);
// Set the file global sequence number to `seqno`
Status AssignGlobalSeqnoForIngestedFile(IngestedFileInfo* file_to_ingest,
SequenceNumber seqno);
// Generate the file checksum and store in the IngestedFileInfo
IOStatus GenerateChecksumForIngestedFile(IngestedFileInfo* file_to_ingest);
// Check if `file_to_ingest` can fit in level `level`
// REQUIRES: Mutex held
bool IngestedFileFitInLevel(const IngestedFileInfo* file_to_ingest,
int level);
// Helper method to sync given file.
template <typename TWritableFile>
Status SyncIngestedFile(TWritableFile* file);
SystemClock* clock_;
FileSystemPtr fs_;
VersionSet* versions_;
ColumnFamilyData* cfd_;
const ImmutableDBOptions& db_options_;
const EnvOptions& env_options_;
SnapshotList* db_snapshots_;
autovector<IngestedFileInfo> files_to_ingest_;
const IngestExternalFileOptions& ingestion_options_;
Directories* directories_;
EventLogger* event_logger_;
VersionEdit edit_;
uint64_t job_start_time_;
int consumed_seqno_count_;
// Set in ExternalSstFileIngestionJob::Prepare(), if true all files are
// ingested in L0
bool files_overlap_{false};
// Set in ExternalSstFileIngestionJob::Prepare(), if true and DB
// file_checksum_gen_factory is set, DB will generate checksum each file.
bool need_generate_file_checksum_{true};
std::shared_ptr<IOTracer> io_tracer_;
};
} // namespace ROCKSDB_NAMESPACE