rocksdb/db/dbformat.cc
Yanqin Jin ae152ee666 Avoid user key copying for Get/Put/Write with user-timestamp (#5502)
Summary:
In previous https://github.com/facebook/rocksdb/issues/5079, we added user-specified timestamp to `DB::Get()` and `DB::Put()`. Limitation is that these two functions may cause extra memory allocation and key copy. The reason is that `WriteBatch` does not allocate extra memory for timestamps because it is not aware of timestamp size, and we did not provide an API to assign/update timestamp of each key within a `WriteBatch`.
We address these issues in this PR by doing the following.
1. Add a `timestamp_size_` to `WriteBatch` so that `WriteBatch` can take timestamps into account when calling `WriteBatch::Put`, `WriteBatch::Delete`, etc.
2. Add APIs `WriteBatch::AssignTimestamp` and `WriteBatch::AssignTimestamps` so that application can assign/update timestamps for each key in a `WriteBatch`.
3. Avoid key copy in `GetImpl` by adding new constructor to `LookupKey`.

Test plan (on devserver):
```
$make clean && COMPILE_WITH_ASAN=1 make -j32 all
$./db_basic_test --gtest_filter=Timestamp/DBBasicTestWithTimestampWithParam.PutAndGet/*
$make check
```
If the API extension looks good, I will add more unit tests.

Some simple benchmark using db_bench.
```
$rm -rf /dev/shm/dbbench/* && TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=fillseq,readrandom -num=1000000
$rm -rf /dev/shm/dbbench/* && TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=fillrandom -num=1000000 -disable_wal=true
```
Master is at a78503bd6c.
```
|        | readrandom | fillrandom |
| master | 15.53 MB/s | 25.97 MB/s |
| PR5502 | 16.70 MB/s | 25.80 MB/s |
```
Pull Request resolved: https://github.com/facebook/rocksdb/pull/5502

Differential Revision: D16340894

Pulled By: riversand963

fbshipit-source-id: 51132cf792be07d1efc3ac33f5768c4ee2608bb8
2019-07-25 15:27:39 -07:00

198 lines
6.3 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).
//
// 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.
#include "db/dbformat.h"
#include <cinttypes>
#include <stdio.h>
#include "monitoring/perf_context_imp.h"
#include "port/port.h"
#include "util/coding.h"
#include "util/string_util.h"
namespace rocksdb {
// kValueTypeForSeek defines the ValueType that should be passed when
// constructing a ParsedInternalKey object for seeking to a particular
// sequence number (since we sort sequence numbers in decreasing order
// and the value type is embedded as the low 8 bits in the sequence
// number in internal keys, we need to use the highest-numbered
// ValueType, not the lowest).
const ValueType kValueTypeForSeek = kTypeBlobIndex;
const ValueType kValueTypeForSeekForPrev = kTypeDeletion;
uint64_t PackSequenceAndType(uint64_t seq, ValueType t) {
assert(seq <= kMaxSequenceNumber);
assert(IsExtendedValueType(t));
return (seq << 8) | t;
}
EntryType GetEntryType(ValueType value_type) {
switch (value_type) {
case kTypeValue:
return kEntryPut;
case kTypeDeletion:
return kEntryDelete;
case kTypeSingleDeletion:
return kEntrySingleDelete;
case kTypeMerge:
return kEntryMerge;
case kTypeRangeDeletion:
return kEntryRangeDeletion;
case kTypeBlobIndex:
return kEntryBlobIndex;
default:
return kEntryOther;
}
}
bool ParseFullKey(const Slice& internal_key, FullKey* fkey) {
ParsedInternalKey ikey;
if (!ParseInternalKey(internal_key, &ikey)) {
return false;
}
fkey->user_key = ikey.user_key;
fkey->sequence = ikey.sequence;
fkey->type = GetEntryType(ikey.type);
return true;
}
void UnPackSequenceAndType(uint64_t packed, uint64_t* seq, ValueType* t) {
*seq = packed >> 8;
*t = static_cast<ValueType>(packed & 0xff);
assert(*seq <= kMaxSequenceNumber);
assert(IsExtendedValueType(*t));
}
void AppendInternalKey(std::string* result, const ParsedInternalKey& key) {
result->append(key.user_key.data(), key.user_key.size());
PutFixed64(result, PackSequenceAndType(key.sequence, key.type));
}
void AppendInternalKeyFooter(std::string* result, SequenceNumber s,
ValueType t) {
PutFixed64(result, PackSequenceAndType(s, t));
}
std::string ParsedInternalKey::DebugString(bool hex) const {
char buf[50];
snprintf(buf, sizeof(buf), "' seq:%" PRIu64 ", type:%d", sequence,
static_cast<int>(type));
std::string result = "'";
result += user_key.ToString(hex);
result += buf;
return result;
}
std::string InternalKey::DebugString(bool hex) const {
std::string result;
ParsedInternalKey parsed;
if (ParseInternalKey(rep_, &parsed)) {
result = parsed.DebugString(hex);
} else {
result = "(bad)";
result.append(EscapeString(rep_));
}
return result;
}
const char* InternalKeyComparator::Name() const { return name_.c_str(); }
int InternalKeyComparator::Compare(const ParsedInternalKey& a,
const ParsedInternalKey& b) const {
// Order by:
// increasing user key (according to user-supplied comparator)
// decreasing sequence number
// decreasing type (though sequence# should be enough to disambiguate)
int r = user_comparator_.Compare(a.user_key, b.user_key);
if (r == 0) {
if (a.sequence > b.sequence) {
r = -1;
} else if (a.sequence < b.sequence) {
r = +1;
} else if (a.type > b.type) {
r = -1;
} else if (a.type < b.type) {
r = +1;
}
}
return r;
}
void InternalKeyComparator::FindShortestSeparator(std::string* start,
const Slice& limit) const {
// Attempt to shorten the user portion of the key
Slice user_start = ExtractUserKey(*start);
Slice user_limit = ExtractUserKey(limit);
std::string tmp(user_start.data(), user_start.size());
user_comparator_.FindShortestSeparator(&tmp, user_limit);
if (tmp.size() <= user_start.size() &&
user_comparator_.Compare(user_start, tmp) < 0) {
// User key has become shorter physically, but larger logically.
// Tack on the earliest possible number to the shortened user key.
PutFixed64(&tmp,
PackSequenceAndType(kMaxSequenceNumber, kValueTypeForSeek));
assert(this->Compare(*start, tmp) < 0);
assert(this->Compare(tmp, limit) < 0);
start->swap(tmp);
}
}
void InternalKeyComparator::FindShortSuccessor(std::string* key) const {
Slice user_key = ExtractUserKey(*key);
std::string tmp(user_key.data(), user_key.size());
user_comparator_.FindShortSuccessor(&tmp);
if (tmp.size() <= user_key.size() &&
user_comparator_.Compare(user_key, tmp) < 0) {
// User key has become shorter physically, but larger logically.
// Tack on the earliest possible number to the shortened user key.
PutFixed64(&tmp,
PackSequenceAndType(kMaxSequenceNumber, kValueTypeForSeek));
assert(this->Compare(*key, tmp) < 0);
key->swap(tmp);
}
}
LookupKey::LookupKey(const Slice& _user_key, SequenceNumber s,
const Slice* ts) {
size_t usize = _user_key.size();
size_t ts_sz = (nullptr == ts) ? 0 : ts->size();
size_t needed = usize + ts_sz + 13; // A conservative estimate
char* dst;
if (needed <= sizeof(space_)) {
dst = space_;
} else {
dst = new char[needed];
}
start_ = dst;
// NOTE: We don't support users keys of more than 2GB :)
dst = EncodeVarint32(dst, static_cast<uint32_t>(usize + ts_sz + 8));
kstart_ = dst;
memcpy(dst, _user_key.data(), usize);
dst += usize;
if (nullptr != ts) {
memcpy(dst, ts->data(), ts_sz);
dst += ts_sz;
}
EncodeFixed64(dst, PackSequenceAndType(s, kValueTypeForSeek));
dst += 8;
end_ = dst;
}
void IterKey::EnlargeBuffer(size_t key_size) {
// If size is smaller than buffer size, continue using current buffer,
// or the static allocated one, as default
assert(key_size > buf_size_);
// Need to enlarge the buffer.
ResetBuffer();
buf_ = new char[key_size];
buf_size_ = key_size;
}
} // namespace rocksdb