rocksdb/db/dbformat.h
Dhruba Borthakur 47c4191fe8 Reduce write amplification by merging files in L0 back into L0
Summary:
There is a new option called hybrid_mode which, when switched on,
causes HBase style compactions.  Files from L0 are
compacted back into L0. This meat of this compaction algorithm
is in PickCompactionHybrid().

All files reside in L0. That means all files have overlapping
keys. Each file has a time-bound, i.e. each file contains a
range of keys that were inserted around the same time. The
start-seqno and the end-seqno refers to the timeframe when
these keys were inserted.  Files that have contiguous seqno
are compacted together into a larger file. All files are
ordered from most recent to the oldest.

The current compaction algorithm starts to look for
candidate files starting from the most recent file. It continues to
add more files to the same compaction run as long as the
sum of the files chosen till now is smaller than the next
candidate file size. This logic needs to be debated
and validated.

The above logic should reduce write amplification to a
large extent... will publish numbers shortly.

Test Plan: dbstress runs for 6 hours with no data corruption (tested so far).

Differential Revision: https://reviews.facebook.net/D11289
2013-06-30 20:07:04 -07:00

228 lines
7.5 KiB
C++

// 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.
#ifndef STORAGE_LEVELDB_DB_FORMAT_H_
#define STORAGE_LEVELDB_DB_FORMAT_H_
#include <stdio.h>
#include "leveldb/comparator.h"
#include "leveldb/db.h"
#include "leveldb/filter_policy.h"
#include "leveldb/slice.h"
#include "leveldb/table_builder.h"
#include "leveldb/types.h"
#include "util/coding.h"
#include "util/logging.h"
namespace leveldb {
class InternalKey;
// Value types encoded as the last component of internal keys.
// DO NOT CHANGE THESE ENUM VALUES: they are embedded in the on-disk
// data structures.
enum ValueType {
kTypeDeletion = 0x0,
kTypeValue = 0x1,
kTypeMerge = 0x2
};
// 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).
static const ValueType kValueTypeForSeek = kTypeMerge;
// We leave eight bits empty at the bottom so a type and sequence#
// can be packed together into 64-bits.
static const SequenceNumber kMaxSequenceNumber =
((0x1ull << 56) - 1);
struct ParsedInternalKey {
Slice user_key;
SequenceNumber sequence;
ValueType type;
ParsedInternalKey() { } // Intentionally left uninitialized (for speed)
ParsedInternalKey(const Slice& u, const SequenceNumber& seq, ValueType t)
: user_key(u), sequence(seq), type(t) { }
std::string DebugString(bool hex = false) const;
};
// Return the length of the encoding of "key".
inline size_t InternalKeyEncodingLength(const ParsedInternalKey& key) {
return key.user_key.size() + 8;
}
// Append the serialization of "key" to *result.
extern void AppendInternalKey(std::string* result,
const ParsedInternalKey& key);
// Attempt to parse an internal key from "internal_key". On success,
// stores the parsed data in "*result", and returns true.
//
// On error, returns false, leaves "*result" in an undefined state.
extern bool ParseInternalKey(const Slice& internal_key,
ParsedInternalKey* result);
// Returns the user key portion of an internal key.
inline Slice ExtractUserKey(const Slice& internal_key) {
assert(internal_key.size() >= 8);
return Slice(internal_key.data(), internal_key.size() - 8);
}
inline ValueType ExtractValueType(const Slice& internal_key) {
assert(internal_key.size() >= 8);
const size_t n = internal_key.size();
uint64_t num = DecodeFixed64(internal_key.data() + n - 8);
unsigned char c = num & 0xff;
return static_cast<ValueType>(c);
}
// A comparator for internal keys that uses a specified comparator for
// the user key portion and breaks ties by decreasing sequence number.
class InternalKeyComparator : public Comparator {
private:
const Comparator* user_comparator_;
std::string name_;
public:
explicit InternalKeyComparator(const Comparator* c) : user_comparator_(c),
name_("leveldb.InternalKeyComparator:" +
std::string(user_comparator_->Name())) {
}
virtual const char* Name() const;
virtual int Compare(const Slice& a, const Slice& b) const;
virtual void FindShortestSeparator(
std::string* start,
const Slice& limit) const;
virtual void FindShortSuccessor(std::string* key) const;
const Comparator* user_comparator() const { return user_comparator_; }
int Compare(const InternalKey& a, const InternalKey& b) const;
};
// Filter policy wrapper that converts from internal keys to user keys
class InternalFilterPolicy : public FilterPolicy {
private:
const FilterPolicy* const user_policy_;
public:
explicit InternalFilterPolicy(const FilterPolicy* p) : user_policy_(p) { }
virtual const char* Name() const;
virtual void CreateFilter(const Slice* keys, int n, std::string* dst) const;
virtual bool KeyMayMatch(const Slice& key, const Slice& filter) const;
};
// Modules in this directory should keep internal keys wrapped inside
// the following class instead of plain strings so that we do not
// incorrectly use string comparisons instead of an InternalKeyComparator.
class InternalKey {
private:
std::string rep_;
public:
InternalKey() { } // Leave rep_ as empty to indicate it is invalid
InternalKey(const Slice& user_key, SequenceNumber s, ValueType t) {
AppendInternalKey(&rep_, ParsedInternalKey(user_key, s, t));
}
void DecodeFrom(const Slice& s) { rep_.assign(s.data(), s.size()); }
Slice Encode() const {
assert(!rep_.empty());
return rep_;
}
Slice user_key() const { return ExtractUserKey(rep_); }
void SetFrom(const ParsedInternalKey& p) {
rep_.clear();
AppendInternalKey(&rep_, p);
}
void Clear() { rep_.clear(); }
std::string DebugString(bool hex = false) const;
};
inline int InternalKeyComparator::Compare(
const InternalKey& a, const InternalKey& b) const {
return Compare(a.Encode(), b.Encode());
}
inline bool ParseInternalKey(const Slice& internal_key,
ParsedInternalKey* result) {
const size_t n = internal_key.size();
if (n < 8) return false;
uint64_t num = DecodeFixed64(internal_key.data() + n - 8);
unsigned char c = num & 0xff;
result->sequence = num >> 8;
result->type = static_cast<ValueType>(c);
result->user_key = Slice(internal_key.data(), n - 8);
return (c <= static_cast<unsigned char>(kValueTypeForSeek));
}
// Update the sequence number in the internal key
inline void UpdateInternalKey(char* internal_key,
const size_t internal_key_size,
uint64_t seq, ValueType t) {
assert(internal_key_size >= 8);
char* seqtype = internal_key + internal_key_size - 8;
uint64_t newval = (seq << 8) | t;
EncodeFixed64(seqtype, newval);
}
// Get the sequence number from the internal key
inline uint64_t GetInternalKeySeqno(const Slice& internal_key) {
const size_t n = internal_key.size();
assert(n >= 8);
uint64_t num = DecodeFixed64(internal_key.data() + n - 8);
return num >> 8;
}
// A helper class useful for DBImpl::Get()
class LookupKey {
public:
// Initialize *this for looking up user_key at a snapshot with
// the specified sequence number.
LookupKey(const Slice& user_key, SequenceNumber sequence);
~LookupKey();
// Return a key suitable for lookup in a MemTable.
Slice memtable_key() const { return Slice(start_, end_ - start_); }
// Return an internal key (suitable for passing to an internal iterator)
Slice internal_key() const { return Slice(kstart_, end_ - kstart_); }
// Return the user key
Slice user_key() const { return Slice(kstart_, end_ - kstart_ - 8); }
private:
// We construct a char array of the form:
// klength varint32 <-- start_
// userkey char[klength] <-- kstart_
// tag uint64
// <-- end_
// The array is a suitable MemTable key.
// The suffix starting with "userkey" can be used as an InternalKey.
const char* start_;
const char* kstart_;
const char* end_;
char space_[200]; // Avoid allocation for short keys
// No copying allowed
LookupKey(const LookupKey&);
void operator=(const LookupKey&);
};
inline LookupKey::~LookupKey() {
if (start_ != space_) delete[] start_;
}
} // namespace leveldb
#endif // STORAGE_LEVELDB_DB_FORMAT_H_