|
|
|
// 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).
|
|
|
|
//
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
// 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.
|
|
|
|
//
|
|
|
|
// InlineSkipList is derived from SkipList (skiplist.h), but it optimizes
|
|
|
|
// the memory layout by requiring that the key storage be allocated through
|
|
|
|
// the skip list instance. For the common case of SkipList<const char*,
|
|
|
|
// Cmp> this saves 1 pointer per skip list node and gives better cache
|
|
|
|
// locality, at the expense of wasted padding from using AllocateAligned
|
|
|
|
// instead of Allocate for the keys. The unused padding will be from
|
|
|
|
// 0 to sizeof(void*)-1 bytes, and the space savings are sizeof(void*)
|
|
|
|
// bytes, so despite the padding the space used is always less than
|
|
|
|
// SkipList<const char*, ..>.
|
|
|
|
//
|
|
|
|
// Thread safety -------------
|
|
|
|
//
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
// Writes via Insert require external synchronization, most likely a mutex.
|
|
|
|
// InsertConcurrently can be safely called concurrently with reads and
|
|
|
|
// with other concurrent inserts. Reads require a guarantee that the
|
|
|
|
// InlineSkipList will not be destroyed while the read is in progress.
|
|
|
|
// Apart from that, reads progress without any internal locking or
|
|
|
|
// synchronization.
|
|
|
|
//
|
|
|
|
// Invariants:
|
|
|
|
//
|
|
|
|
// (1) Allocated nodes are never deleted until the InlineSkipList is
|
|
|
|
// destroyed. This is trivially guaranteed by the code since we never
|
|
|
|
// delete any skip list nodes.
|
|
|
|
//
|
|
|
|
// (2) The contents of a Node except for the next/prev pointers are
|
|
|
|
// immutable after the Node has been linked into the InlineSkipList.
|
|
|
|
// Only Insert() modifies the list, and it is careful to initialize a
|
|
|
|
// node and use release-stores to publish the nodes in one or more lists.
|
|
|
|
//
|
|
|
|
// ... prev vs. next pointer ordering ...
|
|
|
|
//
|
|
|
|
|
|
|
|
#pragma once
|
|
|
|
#include <assert.h>
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <algorithm>
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
#include <atomic>
|
|
|
|
#include "port/likely.h"
|
|
|
|
#include "port/port.h"
|
|
|
|
#include "util/allocator.h"
|
|
|
|
#include "util/random.h"
|
|
|
|
|
|
|
|
namespace rocksdb {
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
class InlineSkipList {
|
|
|
|
private:
|
|
|
|
struct Node;
|
|
|
|
struct Splice;
|
|
|
|
|
|
|
|
public:
|
|
|
|
static const uint16_t kMaxPossibleHeight = 32;
|
|
|
|
|
|
|
|
// Create a new InlineSkipList object that will use "cmp" for comparing
|
|
|
|
// keys, and will allocate memory using "*allocator". Objects allocated
|
|
|
|
// in the allocator must remain allocated for the lifetime of the
|
|
|
|
// skiplist object.
|
|
|
|
explicit InlineSkipList(Comparator cmp, Allocator* allocator,
|
|
|
|
int32_t max_height = 12,
|
|
|
|
int32_t branching_factor = 4);
|
|
|
|
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
// Allocates a key and a skip-list node, returning a pointer to the key
|
|
|
|
// portion of the node. This method is thread-safe if the allocator
|
|
|
|
// is thread-safe.
|
|
|
|
char* AllocateKey(size_t key_size);
|
|
|
|
|
|
|
|
// Allocate a splice using allocator.
|
|
|
|
Splice* AllocateSplice();
|
|
|
|
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
// Inserts a key allocated by AllocateKey, after the actual key value
|
|
|
|
// has been filled in.
|
|
|
|
//
|
|
|
|
// REQUIRES: nothing that compares equal to key is currently in the list.
|
|
|
|
// REQUIRES: no concurrent calls to any of inserts.
|
|
|
|
bool Insert(const char* key);
|
|
|
|
|
|
|
|
// Inserts a key allocated by AllocateKey with a hint of last insert
|
|
|
|
// position in the skip-list. If hint points to nullptr, a new hint will be
|
|
|
|
// populated, which can be used in subsequent calls.
|
|
|
|
//
|
|
|
|
// It can be used to optimize the workload where there are multiple groups
|
|
|
|
// of keys, and each key is likely to insert to a location close to the last
|
|
|
|
// inserted key in the same group. One example is sequential inserts.
|
|
|
|
//
|
|
|
|
// REQUIRES: nothing that compares equal to key is currently in the list.
|
|
|
|
// REQUIRES: no concurrent calls to any of inserts.
|
|
|
|
bool InsertWithHint(const char* key, void** hint);
|
|
|
|
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
// Like Insert, but external synchronization is not required.
|
|
|
|
bool InsertConcurrently(const char* key);
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
|
|
|
|
// Inserts a node into the skip list. key must have been allocated by
|
|
|
|
// AllocateKey and then filled in by the caller. If UseCAS is true,
|
|
|
|
// then external synchronization is not required, otherwise this method
|
|
|
|
// may not be called concurrently with any other insertions.
|
|
|
|
//
|
|
|
|
// Regardless of whether UseCAS is true, the splice must be owned
|
|
|
|
// exclusively by the current thread. If allow_partial_splice_fix is
|
|
|
|
// true, then the cost of insertion is amortized O(log D), where D is
|
|
|
|
// the distance from the splice to the inserted key (measured as the
|
|
|
|
// number of intervening nodes). Note that this bound is very good for
|
|
|
|
// sequential insertions! If allow_partial_splice_fix is false then
|
|
|
|
// the existing splice will be ignored unless the current key is being
|
|
|
|
// inserted immediately after the splice. allow_partial_splice_fix ==
|
|
|
|
// false has worse running time for the non-sequential case O(log N),
|
|
|
|
// but a better constant factor.
|
|
|
|
template <bool UseCAS>
|
|
|
|
bool Insert(const char* key, Splice* splice, bool allow_partial_splice_fix);
|
|
|
|
|
|
|
|
// Returns true iff an entry that compares equal to key is in the list.
|
|
|
|
bool Contains(const char* key) const;
|
|
|
|
|
|
|
|
// Return estimated number of entries smaller than `key`.
|
|
|
|
uint64_t EstimateCount(const char* key) const;
|
|
|
|
|
|
|
|
// Validate correctness of the skip-list.
|
|
|
|
void TEST_Validate() const;
|
|
|
|
|
|
|
|
// Iteration over the contents of a skip list
|
|
|
|
class Iterator {
|
|
|
|
public:
|
|
|
|
// Initialize an iterator over the specified list.
|
|
|
|
// The returned iterator is not valid.
|
|
|
|
explicit Iterator(const InlineSkipList* list);
|
|
|
|
|
|
|
|
// Change the underlying skiplist used for this iterator
|
|
|
|
// This enables us not changing the iterator without deallocating
|
|
|
|
// an old one and then allocating a new one
|
|
|
|
void SetList(const InlineSkipList* list);
|
|
|
|
|
|
|
|
// Returns true iff the iterator is positioned at a valid node.
|
|
|
|
bool Valid() const;
|
|
|
|
|
|
|
|
// Returns the key at the current position.
|
|
|
|
// REQUIRES: Valid()
|
|
|
|
const char* key() const;
|
|
|
|
|
|
|
|
// Advances to the next position.
|
|
|
|
// REQUIRES: Valid()
|
|
|
|
void Next();
|
|
|
|
|
|
|
|
// Advances to the previous position.
|
|
|
|
// REQUIRES: Valid()
|
|
|
|
void Prev();
|
|
|
|
|
|
|
|
// Advance to the first entry with a key >= target
|
|
|
|
void Seek(const char* target);
|
|
|
|
|
|
|
|
// Retreat to the last entry with a key <= target
|
|
|
|
void SeekForPrev(const char* target);
|
|
|
|
|
|
|
|
// Position at the first entry in list.
|
|
|
|
// Final state of iterator is Valid() iff list is not empty.
|
|
|
|
void SeekToFirst();
|
|
|
|
|
|
|
|
// Position at the last entry in list.
|
|
|
|
// Final state of iterator is Valid() iff list is not empty.
|
|
|
|
void SeekToLast();
|
|
|
|
|
|
|
|
private:
|
|
|
|
const InlineSkipList* list_;
|
|
|
|
Node* node_;
|
|
|
|
// Intentionally copyable
|
|
|
|
};
|
|
|
|
|
|
|
|
private:
|
|
|
|
const uint16_t kMaxHeight_;
|
|
|
|
const uint16_t kBranching_;
|
|
|
|
const uint32_t kScaledInverseBranching_;
|
|
|
|
|
|
|
|
// Immutable after construction
|
|
|
|
Comparator const compare_;
|
|
|
|
Allocator* const allocator_; // Allocator used for allocations of nodes
|
|
|
|
|
|
|
|
Node* const head_;
|
|
|
|
|
|
|
|
// Modified only by Insert(). Read racily by readers, but stale
|
|
|
|
// values are ok.
|
|
|
|
std::atomic<int> max_height_; // Height of the entire list
|
|
|
|
|
|
|
|
// seq_splice_ is a Splice used for insertions in the non-concurrent
|
|
|
|
// case. It caches the prev and next found during the most recent
|
|
|
|
// non-concurrent insertion.
|
|
|
|
Splice* seq_splice_;
|
|
|
|
|
|
|
|
inline int GetMaxHeight() const {
|
|
|
|
return max_height_.load(std::memory_order_relaxed);
|
|
|
|
}
|
|
|
|
|
|
|
|
int RandomHeight();
|
|
|
|
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
Node* AllocateNode(size_t key_size, int height);
|
|
|
|
|
|
|
|
bool Equal(const char* a, const char* b) const {
|
|
|
|
return (compare_(a, b) == 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
bool LessThan(const char* a, const char* b) const {
|
|
|
|
return (compare_(a, b) < 0);
|
|
|
|
}
|
|
|
|
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
// Return true if key is greater than the data stored in "n". Null n
|
|
|
|
// is considered infinite. n should not be head_.
|
|
|
|
bool KeyIsAfterNode(const char* key, Node* n) const;
|
|
|
|
|
|
|
|
// Returns the earliest node with a key >= key.
|
|
|
|
// Return nullptr if there is no such node.
|
|
|
|
Node* FindGreaterOrEqual(const char* key) const;
|
|
|
|
|
|
|
|
// Return the latest node with a key < key.
|
|
|
|
// Return head_ if there is no such node.
|
|
|
|
// Fills prev[level] with pointer to previous node at "level" for every
|
|
|
|
// level in [0..max_height_-1], if prev is non-null.
|
|
|
|
Node* FindLessThan(const char* key, Node** prev = nullptr) const;
|
|
|
|
|
|
|
|
// Return the latest node with a key < key on bottom_level. Start searching
|
|
|
|
// from root node on the level below top_level.
|
|
|
|
// Fills prev[level] with pointer to previous node at "level" for every
|
|
|
|
// level in [bottom_level..top_level-1], if prev is non-null.
|
|
|
|
Node* FindLessThan(const char* key, Node** prev, Node* root, int top_level,
|
|
|
|
int bottom_level) const;
|
|
|
|
|
|
|
|
// Return the last node in the list.
|
|
|
|
// Return head_ if list is empty.
|
|
|
|
Node* FindLast() const;
|
|
|
|
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
// Traverses a single level of the list, setting *out_prev to the last
|
|
|
|
// node before the key and *out_next to the first node after. Assumes
|
|
|
|
// that the key is not present in the skip list. On entry, before should
|
|
|
|
// point to a node that is before the key, and after should point to
|
|
|
|
// a node that is after the key. after should be nullptr if a good after
|
|
|
|
// node isn't conveniently available.
|
|
|
|
template<bool prefetch_before>
|
|
|
|
void FindSpliceForLevel(const char* key, Node* before, Node* after, int level,
|
|
|
|
Node** out_prev, Node** out_next);
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
|
|
|
|
// Recomputes Splice levels from highest_level (inclusive) down to
|
|
|
|
// lowest_level (inclusive).
|
|
|
|
void RecomputeSpliceLevels(const char* key, Splice* splice,
|
|
|
|
int recompute_level);
|
|
|
|
|
|
|
|
// No copying allowed
|
|
|
|
InlineSkipList(const InlineSkipList&);
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
InlineSkipList& operator=(const InlineSkipList&);
|
|
|
|
};
|
|
|
|
|
|
|
|
// Implementation details follow
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
struct InlineSkipList<Comparator>::Splice {
|
|
|
|
// The invariant of a Splice is that prev_[i+1].key <= prev_[i].key <
|
|
|
|
// next_[i].key <= next_[i+1].key for all i. That means that if a
|
|
|
|
// key is bracketed by prev_[i] and next_[i] then it is bracketed by
|
|
|
|
// all higher levels. It is _not_ required that prev_[i]->Next(i) ==
|
|
|
|
// next_[i] (it probably did at some point in the past, but intervening
|
|
|
|
// or concurrent operations might have inserted nodes in between).
|
|
|
|
int height_ = 0;
|
|
|
|
Node** prev_;
|
|
|
|
Node** next_;
|
|
|
|
};
|
|
|
|
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
// The Node data type is more of a pointer into custom-managed memory than
|
|
|
|
// a traditional C++ struct. The key is stored in the bytes immediately
|
|
|
|
// after the struct, and the next_ pointers for nodes with height > 1 are
|
|
|
|
// stored immediately _before_ the struct. This avoids the need to include
|
|
|
|
// any pointer or sizing data, which reduces per-node memory overheads.
|
|
|
|
template <class Comparator>
|
|
|
|
struct InlineSkipList<Comparator>::Node {
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
// Stores the height of the node in the memory location normally used for
|
|
|
|
// next_[0]. This is used for passing data from AllocateKey to Insert.
|
|
|
|
void StashHeight(const int height) {
|
|
|
|
assert(sizeof(int) <= sizeof(next_[0]));
|
|
|
|
memcpy(&next_[0], &height, sizeof(int));
|
|
|
|
}
|
|
|
|
|
|
|
|
// Retrieves the value passed to StashHeight. Undefined after a call
|
|
|
|
// to SetNext or NoBarrier_SetNext.
|
|
|
|
int UnstashHeight() const {
|
|
|
|
int rv;
|
|
|
|
memcpy(&rv, &next_[0], sizeof(int));
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
const char* Key() const { return reinterpret_cast<const char*>(&next_[1]); }
|
|
|
|
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
// Accessors/mutators for links. Wrapped in methods so we can add
|
|
|
|
// the appropriate barriers as necessary, and perform the necessary
|
|
|
|
// addressing trickery for storing links below the Node in memory.
|
|
|
|
Node* Next(int n) {
|
|
|
|
assert(n >= 0);
|
|
|
|
// Use an 'acquire load' so that we observe a fully initialized
|
|
|
|
// version of the returned Node.
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
return (next_[-n].load(std::memory_order_acquire));
|
|
|
|
}
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
|
|
|
|
void SetNext(int n, Node* x) {
|
|
|
|
assert(n >= 0);
|
|
|
|
// Use a 'release store' so that anybody who reads through this
|
|
|
|
// pointer observes a fully initialized version of the inserted node.
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
next_[-n].store(x, std::memory_order_release);
|
|
|
|
}
|
|
|
|
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
bool CASNext(int n, Node* expected, Node* x) {
|
|
|
|
assert(n >= 0);
|
|
|
|
return next_[-n].compare_exchange_strong(expected, x);
|
|
|
|
}
|
|
|
|
|
|
|
|
// No-barrier variants that can be safely used in a few locations.
|
|
|
|
Node* NoBarrier_Next(int n) {
|
|
|
|
assert(n >= 0);
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
return next_[-n].load(std::memory_order_relaxed);
|
|
|
|
}
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
|
|
|
|
void NoBarrier_SetNext(int n, Node* x) {
|
|
|
|
assert(n >= 0);
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
next_[-n].store(x, std::memory_order_relaxed);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Insert node after prev on specific level.
|
|
|
|
void InsertAfter(Node* prev, int level) {
|
|
|
|
// NoBarrier_SetNext() suffices since we will add a barrier when
|
|
|
|
// we publish a pointer to "this" in prev.
|
|
|
|
NoBarrier_SetNext(level, prev->NoBarrier_Next(level));
|
|
|
|
prev->SetNext(level, this);
|
|
|
|
}
|
|
|
|
|
|
|
|
private:
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
// next_[0] is the lowest level link (level 0). Higher levels are
|
|
|
|
// stored _earlier_, so level 1 is at next_[-1].
|
|
|
|
std::atomic<Node*> next_[1];
|
|
|
|
};
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
inline InlineSkipList<Comparator>::Iterator::Iterator(
|
|
|
|
const InlineSkipList* list) {
|
|
|
|
SetList(list);
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
inline void InlineSkipList<Comparator>::Iterator::SetList(
|
|
|
|
const InlineSkipList* list) {
|
|
|
|
list_ = list;
|
|
|
|
node_ = nullptr;
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
inline bool InlineSkipList<Comparator>::Iterator::Valid() const {
|
|
|
|
return node_ != nullptr;
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
inline const char* InlineSkipList<Comparator>::Iterator::key() const {
|
|
|
|
assert(Valid());
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
return node_->Key();
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
inline void InlineSkipList<Comparator>::Iterator::Next() {
|
|
|
|
assert(Valid());
|
|
|
|
node_ = node_->Next(0);
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
inline void InlineSkipList<Comparator>::Iterator::Prev() {
|
|
|
|
// Instead of using explicit "prev" links, we just search for the
|
|
|
|
// last node that falls before key.
|
|
|
|
assert(Valid());
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
node_ = list_->FindLessThan(node_->Key());
|
|
|
|
if (node_ == list_->head_) {
|
|
|
|
node_ = nullptr;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
inline void InlineSkipList<Comparator>::Iterator::Seek(const char* target) {
|
|
|
|
node_ = list_->FindGreaterOrEqual(target);
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
inline void InlineSkipList<Comparator>::Iterator::SeekForPrev(
|
|
|
|
const char* target) {
|
|
|
|
Seek(target);
|
|
|
|
if (!Valid()) {
|
|
|
|
SeekToLast();
|
|
|
|
}
|
|
|
|
while (Valid() && list_->LessThan(target, key())) {
|
|
|
|
Prev();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
inline void InlineSkipList<Comparator>::Iterator::SeekToFirst() {
|
|
|
|
node_ = list_->head_->Next(0);
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
inline void InlineSkipList<Comparator>::Iterator::SeekToLast() {
|
|
|
|
node_ = list_->FindLast();
|
|
|
|
if (node_ == list_->head_) {
|
|
|
|
node_ = nullptr;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
int InlineSkipList<Comparator>::RandomHeight() {
|
|
|
|
auto rnd = Random::GetTLSInstance();
|
|
|
|
|
|
|
|
// Increase height with probability 1 in kBranching
|
|
|
|
int height = 1;
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
while (height < kMaxHeight_ && height < kMaxPossibleHeight &&
|
|
|
|
rnd->Next() < kScaledInverseBranching_) {
|
|
|
|
height++;
|
|
|
|
}
|
|
|
|
assert(height > 0);
|
|
|
|
assert(height <= kMaxHeight_);
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
assert(height <= kMaxPossibleHeight);
|
|
|
|
return height;
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
bool InlineSkipList<Comparator>::KeyIsAfterNode(const char* key,
|
|
|
|
Node* n) const {
|
|
|
|
// nullptr n is considered infinite
|
|
|
|
assert(n != head_);
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
return (n != nullptr) && (compare_(n->Key(), key) < 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
typename InlineSkipList<Comparator>::Node*
|
|
|
|
InlineSkipList<Comparator>::FindGreaterOrEqual(const char* key) const {
|
|
|
|
// Note: It looks like we could reduce duplication by implementing
|
|
|
|
// this function as FindLessThan(key)->Next(0), but we wouldn't be able
|
|
|
|
// to exit early on equality and the result wouldn't even be correct.
|
|
|
|
// A concurrent insert might occur after FindLessThan(key) but before
|
|
|
|
// we get a chance to call Next(0).
|
|
|
|
Node* x = head_;
|
|
|
|
int level = GetMaxHeight() - 1;
|
|
|
|
Node* last_bigger = nullptr;
|
|
|
|
while (true) {
|
|
|
|
Node* next = x->Next(level);
|
|
|
|
if (next != nullptr) {
|
|
|
|
PREFETCH(next->Next(level), 0, 1);
|
|
|
|
}
|
|
|
|
// Make sure the lists are sorted
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
assert(x == head_ || next == nullptr || KeyIsAfterNode(next->Key(), x));
|
|
|
|
// Make sure we haven't overshot during our search
|
|
|
|
assert(x == head_ || KeyIsAfterNode(key, x));
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
int cmp = (next == nullptr || next == last_bigger)
|
|
|
|
? 1
|
|
|
|
: compare_(next->Key(), key);
|
|
|
|
if (cmp == 0 || (cmp > 0 && level == 0)) {
|
|
|
|
return next;
|
|
|
|
} else if (cmp < 0) {
|
|
|
|
// Keep searching in this list
|
|
|
|
x = next;
|
|
|
|
} else {
|
|
|
|
// Switch to next list, reuse compare_() result
|
|
|
|
last_bigger = next;
|
|
|
|
level--;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
typename InlineSkipList<Comparator>::Node*
|
|
|
|
InlineSkipList<Comparator>::FindLessThan(const char* key, Node** prev) const {
|
|
|
|
return FindLessThan(key, prev, head_, GetMaxHeight(), 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
typename InlineSkipList<Comparator>::Node*
|
|
|
|
InlineSkipList<Comparator>::FindLessThan(const char* key, Node** prev,
|
|
|
|
Node* root, int top_level,
|
|
|
|
int bottom_level) const {
|
|
|
|
assert(top_level > bottom_level);
|
|
|
|
int level = top_level - 1;
|
|
|
|
Node* x = root;
|
|
|
|
// KeyIsAfter(key, last_not_after) is definitely false
|
|
|
|
Node* last_not_after = nullptr;
|
|
|
|
while (true) {
|
|
|
|
assert(x != nullptr);
|
|
|
|
Node* next = x->Next(level);
|
|
|
|
if (next != nullptr) {
|
|
|
|
PREFETCH(next->Next(level), 0, 1);
|
|
|
|
}
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
assert(x == head_ || next == nullptr || KeyIsAfterNode(next->Key(), x));
|
|
|
|
assert(x == head_ || KeyIsAfterNode(key, x));
|
|
|
|
if (next != last_not_after && KeyIsAfterNode(key, next)) {
|
|
|
|
// Keep searching in this list
|
|
|
|
assert(next != nullptr);
|
|
|
|
x = next;
|
|
|
|
} else {
|
|
|
|
if (prev != nullptr) {
|
|
|
|
prev[level] = x;
|
|
|
|
}
|
|
|
|
if (level == bottom_level) {
|
|
|
|
return x;
|
|
|
|
} else {
|
|
|
|
// Switch to next list, reuse KeyIsAfterNode() result
|
|
|
|
last_not_after = next;
|
|
|
|
level--;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
typename InlineSkipList<Comparator>::Node*
|
|
|
|
InlineSkipList<Comparator>::FindLast() const {
|
|
|
|
Node* x = head_;
|
|
|
|
int level = GetMaxHeight() - 1;
|
|
|
|
while (true) {
|
|
|
|
Node* next = x->Next(level);
|
|
|
|
if (next == nullptr) {
|
|
|
|
if (level == 0) {
|
|
|
|
return x;
|
|
|
|
} else {
|
|
|
|
// Switch to next list
|
|
|
|
level--;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
x = next;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
uint64_t InlineSkipList<Comparator>::EstimateCount(const char* key) const {
|
|
|
|
uint64_t count = 0;
|
|
|
|
|
|
|
|
Node* x = head_;
|
|
|
|
int level = GetMaxHeight() - 1;
|
|
|
|
while (true) {
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
assert(x == head_ || compare_(x->Key(), key) < 0);
|
|
|
|
Node* next = x->Next(level);
|
|
|
|
if (next != nullptr) {
|
|
|
|
PREFETCH(next->Next(level), 0, 1);
|
|
|
|
}
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
if (next == nullptr || compare_(next->Key(), key) >= 0) {
|
|
|
|
if (level == 0) {
|
|
|
|
return count;
|
|
|
|
} else {
|
|
|
|
// Switch to next list
|
|
|
|
count *= kBranching_;
|
|
|
|
level--;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
x = next;
|
|
|
|
count++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
InlineSkipList<Comparator>::InlineSkipList(const Comparator cmp,
|
|
|
|
Allocator* allocator,
|
|
|
|
int32_t max_height,
|
|
|
|
int32_t branching_factor)
|
|
|
|
: kMaxHeight_(static_cast<uint16_t>(max_height)),
|
|
|
|
kBranching_(static_cast<uint16_t>(branching_factor)),
|
|
|
|
kScaledInverseBranching_((Random::kMaxNext + 1) / kBranching_),
|
|
|
|
compare_(cmp),
|
|
|
|
allocator_(allocator),
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
head_(AllocateNode(0, max_height)),
|
|
|
|
max_height_(1),
|
|
|
|
seq_splice_(AllocateSplice()) {
|
|
|
|
assert(max_height > 0 && kMaxHeight_ == static_cast<uint32_t>(max_height));
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
assert(branching_factor > 1 &&
|
|
|
|
kBranching_ == static_cast<uint32_t>(branching_factor));
|
|
|
|
assert(kScaledInverseBranching_ > 0);
|
|
|
|
|
|
|
|
for (int i = 0; i < kMaxHeight_; ++i) {
|
|
|
|
head_->SetNext(i, nullptr);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
char* InlineSkipList<Comparator>::AllocateKey(size_t key_size) {
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
return const_cast<char*>(AllocateNode(key_size, RandomHeight())->Key());
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
typename InlineSkipList<Comparator>::Node*
|
|
|
|
InlineSkipList<Comparator>::AllocateNode(size_t key_size, int height) {
|
|
|
|
auto prefix = sizeof(std::atomic<Node*>) * (height - 1);
|
|
|
|
|
|
|
|
// prefix is space for the height - 1 pointers that we store before
|
|
|
|
// the Node instance (next_[-(height - 1) .. -1]). Node starts at
|
|
|
|
// raw + prefix, and holds the bottom-mode (level 0) skip list pointer
|
|
|
|
// next_[0]. key_size is the bytes for the key, which comes just after
|
|
|
|
// the Node.
|
|
|
|
char* raw = allocator_->AllocateAligned(prefix + sizeof(Node) + key_size);
|
|
|
|
Node* x = reinterpret_cast<Node*>(raw + prefix);
|
|
|
|
|
|
|
|
// Once we've linked the node into the skip list we don't actually need
|
|
|
|
// to know its height, because we can implicitly use the fact that we
|
|
|
|
// traversed into a node at level h to known that h is a valid level
|
|
|
|
// for that node. We need to convey the height to the Insert step,
|
|
|
|
// however, so that it can perform the proper links. Since we're not
|
|
|
|
// using the pointers at the moment, StashHeight temporarily borrow
|
|
|
|
// storage from next_[0] for that purpose.
|
|
|
|
x->StashHeight(height);
|
|
|
|
return x;
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
typename InlineSkipList<Comparator>::Splice*
|
|
|
|
InlineSkipList<Comparator>::AllocateSplice() {
|
|
|
|
// size of prev_ and next_
|
|
|
|
size_t array_size = sizeof(Node*) * (kMaxHeight_ + 1);
|
|
|
|
char* raw = allocator_->AllocateAligned(sizeof(Splice) + array_size * 2);
|
|
|
|
Splice* splice = reinterpret_cast<Splice*>(raw);
|
|
|
|
splice->height_ = 0;
|
|
|
|
splice->prev_ = reinterpret_cast<Node**>(raw + sizeof(Splice));
|
|
|
|
splice->next_ = reinterpret_cast<Node**>(raw + sizeof(Splice) + array_size);
|
|
|
|
return splice;
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
bool InlineSkipList<Comparator>::Insert(const char* key) {
|
|
|
|
return Insert<false>(key, seq_splice_, false);
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
bool InlineSkipList<Comparator>::InsertConcurrently(const char* key) {
|
|
|
|
Node* prev[kMaxPossibleHeight];
|
|
|
|
Node* next[kMaxPossibleHeight];
|
|
|
|
Splice splice;
|
|
|
|
splice.prev_ = prev;
|
|
|
|
splice.next_ = next;
|
|
|
|
return Insert<true>(key, &splice, false);
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
bool InlineSkipList<Comparator>::InsertWithHint(const char* key, void** hint) {
|
|
|
|
assert(hint != nullptr);
|
|
|
|
Splice* splice = reinterpret_cast<Splice*>(*hint);
|
|
|
|
if (splice == nullptr) {
|
|
|
|
splice = AllocateSplice();
|
|
|
|
*hint = reinterpret_cast<void*>(splice);
|
|
|
|
}
|
|
|
|
return Insert<false>(key, splice, true);
|
|
|
|
}
|
|
|
|
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
template <class Comparator>
|
|
|
|
template <bool prefetch_before>
|
|
|
|
void InlineSkipList<Comparator>::FindSpliceForLevel(const char* key,
|
|
|
|
Node* before, Node* after,
|
|
|
|
int level, Node** out_prev,
|
|
|
|
Node** out_next) {
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
while (true) {
|
|
|
|
Node* next = before->Next(level);
|
|
|
|
if (next != nullptr) {
|
|
|
|
PREFETCH(next->Next(level), 0, 1);
|
|
|
|
}
|
|
|
|
if (prefetch_before == true) {
|
|
|
|
if (next != nullptr && level>0) {
|
|
|
|
PREFETCH(next->Next(level-1), 0, 1);
|
|
|
|
}
|
|
|
|
}
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
assert(before == head_ || next == nullptr ||
|
|
|
|
KeyIsAfterNode(next->Key(), before));
|
|
|
|
assert(before == head_ || KeyIsAfterNode(key, before));
|
|
|
|
if (next == after || !KeyIsAfterNode(key, next)) {
|
|
|
|
// found it
|
|
|
|
*out_prev = before;
|
|
|
|
*out_next = next;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
before = next;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
void InlineSkipList<Comparator>::RecomputeSpliceLevels(const char* key,
|
|
|
|
Splice* splice,
|
|
|
|
int recompute_level) {
|
|
|
|
assert(recompute_level > 0);
|
|
|
|
assert(recompute_level <= splice->height_);
|
|
|
|
for (int i = recompute_level - 1; i >= 0; --i) {
|
|
|
|
FindSpliceForLevel<true>(key, splice->prev_[i + 1], splice->next_[i + 1], i,
|
|
|
|
&splice->prev_[i], &splice->next_[i]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
template <bool UseCAS>
|
|
|
|
bool InlineSkipList<Comparator>::Insert(const char* key, Splice* splice,
|
|
|
|
bool allow_partial_splice_fix) {
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
Node* x = reinterpret_cast<Node*>(const_cast<char*>(key)) - 1;
|
|
|
|
int height = x->UnstashHeight();
|
|
|
|
assert(height >= 1 && height <= kMaxHeight_);
|
|
|
|
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
int max_height = max_height_.load(std::memory_order_relaxed);
|
|
|
|
while (height > max_height) {
|
|
|
|
if (max_height_.compare_exchange_weak(max_height, height)) {
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
// successfully updated it
|
|
|
|
max_height = height;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
// else retry, possibly exiting the loop because somebody else
|
|
|
|
// increased it
|
|
|
|
}
|
|
|
|
assert(max_height <= kMaxPossibleHeight);
|
|
|
|
|
|
|
|
int recompute_height = 0;
|
|
|
|
if (splice->height_ < max_height) {
|
|
|
|
// Either splice has never been used or max_height has grown since
|
|
|
|
// last use. We could potentially fix it in the latter case, but
|
|
|
|
// that is tricky.
|
|
|
|
splice->prev_[max_height] = head_;
|
|
|
|
splice->next_[max_height] = nullptr;
|
|
|
|
splice->height_ = max_height;
|
|
|
|
recompute_height = max_height;
|
|
|
|
} else {
|
|
|
|
// Splice is a valid proper-height splice that brackets some
|
|
|
|
// key, but does it bracket this one? We need to validate it and
|
|
|
|
// recompute a portion of the splice (levels 0..recompute_height-1)
|
|
|
|
// that is a superset of all levels that don't bracket the new key.
|
|
|
|
// Several choices are reasonable, because we have to balance the work
|
|
|
|
// saved against the extra comparisons required to validate the Splice.
|
|
|
|
//
|
|
|
|
// One strategy is just to recompute all of orig_splice_height if the
|
|
|
|
// bottom level isn't bracketing. This pessimistically assumes that
|
|
|
|
// we will either get a perfect Splice hit (increasing sequential
|
|
|
|
// inserts) or have no locality.
|
|
|
|
//
|
|
|
|
// Another strategy is to walk up the Splice's levels until we find
|
|
|
|
// a level that brackets the key. This strategy lets the Splice
|
|
|
|
// hint help for other cases: it turns insertion from O(log N) into
|
|
|
|
// O(log D), where D is the number of nodes in between the key that
|
|
|
|
// produced the Splice and the current insert (insertion is aided
|
|
|
|
// whether the new key is before or after the splice). If you have
|
|
|
|
// a way of using a prefix of the key to map directly to the closest
|
|
|
|
// Splice out of O(sqrt(N)) Splices and we make it so that splices
|
|
|
|
// can also be used as hints during read, then we end up with Oshman's
|
|
|
|
// and Shavit's SkipTrie, which has O(log log N) lookup and insertion
|
|
|
|
// (compare to O(log N) for skip list).
|
|
|
|
//
|
|
|
|
// We control the pessimistic strategy with allow_partial_splice_fix.
|
|
|
|
// A good strategy is probably to be pessimistic for seq_splice_,
|
|
|
|
// optimistic if the caller actually went to the work of providing
|
|
|
|
// a Splice.
|
|
|
|
while (recompute_height < max_height) {
|
|
|
|
if (splice->prev_[recompute_height]->Next(recompute_height) !=
|
|
|
|
splice->next_[recompute_height]) {
|
|
|
|
// splice isn't tight at this level, there must have been some inserts
|
|
|
|
// to this
|
|
|
|
// location that didn't update the splice. We might only be a little
|
|
|
|
// stale, but if
|
|
|
|
// the splice is very stale it would be O(N) to fix it. We haven't used
|
|
|
|
// up any of
|
|
|
|
// our budget of comparisons, so always move up even if we are
|
|
|
|
// pessimistic about
|
|
|
|
// our chances of success.
|
|
|
|
++recompute_height;
|
|
|
|
} else if (splice->prev_[recompute_height] != head_ &&
|
|
|
|
!KeyIsAfterNode(key, splice->prev_[recompute_height])) {
|
|
|
|
// key is from before splice
|
|
|
|
if (allow_partial_splice_fix) {
|
|
|
|
// skip all levels with the same node without more comparisons
|
|
|
|
Node* bad = splice->prev_[recompute_height];
|
|
|
|
while (splice->prev_[recompute_height] == bad) {
|
|
|
|
++recompute_height;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
// we're pessimistic, recompute everything
|
|
|
|
recompute_height = max_height;
|
|
|
|
}
|
|
|
|
} else if (KeyIsAfterNode(key, splice->next_[recompute_height])) {
|
|
|
|
// key is from after splice
|
|
|
|
if (allow_partial_splice_fix) {
|
|
|
|
Node* bad = splice->next_[recompute_height];
|
|
|
|
while (splice->next_[recompute_height] == bad) {
|
|
|
|
++recompute_height;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
recompute_height = max_height;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
// this level brackets the key, we won!
|
support for concurrent adds to memtable
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
9 years ago
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
assert(recompute_height <= max_height);
|
|
|
|
if (recompute_height > 0) {
|
|
|
|
RecomputeSpliceLevels(key, splice, recompute_height);
|
|
|
|
}
|
|
|
|
|
|
|
|
bool splice_is_valid = true;
|
|
|
|
if (UseCAS) {
|
|
|
|
for (int i = 0; i < height; ++i) {
|
|
|
|
while (true) {
|
|
|
|
// Checking for duplicate keys on the level 0 is sufficient
|
|
|
|
if (UNLIKELY(i == 0 && splice->next_[i] != nullptr &&
|
|
|
|
compare_(x->Key(), splice->next_[i]->Key()) >= 0)) {
|
|
|
|
// duplicate key
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
if (UNLIKELY(i == 0 && splice->prev_[i] != head_ &&
|
|
|
|
compare_(splice->prev_[i]->Key(), x->Key()) >= 0)) {
|
|
|
|
// duplicate key
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
assert(splice->next_[i] == nullptr ||
|
|
|
|
compare_(x->Key(), splice->next_[i]->Key()) < 0);
|
|
|
|
assert(splice->prev_[i] == head_ ||
|
|
|
|
compare_(splice->prev_[i]->Key(), x->Key()) < 0);
|
|
|
|
x->NoBarrier_SetNext(i, splice->next_[i]);
|
|
|
|
if (splice->prev_[i]->CASNext(i, splice->next_[i], x)) {
|
|
|
|
// success
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
// CAS failed, we need to recompute prev and next. It is unlikely
|
|
|
|
// to be helpful to try to use a different level as we redo the
|
|
|
|
// search, because it should be unlikely that lots of nodes have
|
|
|
|
// been inserted between prev[i] and next[i]. No point in using
|
|
|
|
// next[i] as the after hint, because we know it is stale.
|
|
|
|
FindSpliceForLevel<false>(key, splice->prev_[i], nullptr, i, &splice->prev_[i],
|
|
|
|
&splice->next_[i]);
|
|
|
|
|
|
|
|
// Since we've narrowed the bracket for level i, we might have
|
|
|
|
// violated the Splice constraint between i and i-1. Make sure
|
|
|
|
// we recompute the whole thing next time.
|
|
|
|
if (i > 0) {
|
|
|
|
splice_is_valid = false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
for (int i = 0; i < height; ++i) {
|
|
|
|
if (i >= recompute_height &&
|
|
|
|
splice->prev_[i]->Next(i) != splice->next_[i]) {
|
|
|
|
FindSpliceForLevel<false>(key, splice->prev_[i], nullptr, i, &splice->prev_[i],
|
|
|
|
&splice->next_[i]);
|
|
|
|
}
|
|
|
|
// Checking for duplicate keys on the level 0 is sufficient
|
|
|
|
if (UNLIKELY(i == 0 && splice->next_[i] != nullptr &&
|
|
|
|
compare_(x->Key(), splice->next_[i]->Key()) >= 0)) {
|
|
|
|
// duplicate key
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
if (UNLIKELY(i == 0 && splice->prev_[i] != head_ &&
|
|
|
|
compare_(splice->prev_[i]->Key(), x->Key()) >= 0)) {
|
|
|
|
// duplicate key
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
assert(splice->next_[i] == nullptr ||
|
|
|
|
compare_(x->Key(), splice->next_[i]->Key()) < 0);
|
|
|
|
assert(splice->prev_[i] == head_ ||
|
|
|
|
compare_(splice->prev_[i]->Key(), x->Key()) < 0);
|
|
|
|
assert(splice->prev_[i]->Next(i) == splice->next_[i]);
|
|
|
|
x->NoBarrier_SetNext(i, splice->next_[i]);
|
|
|
|
splice->prev_[i]->SetNext(i, x);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (splice_is_valid) {
|
|
|
|
for (int i = 0; i < height; ++i) {
|
|
|
|
splice->prev_[i] = x;
|
|
|
|
}
|
|
|
|
assert(splice->prev_[splice->height_] == head_);
|
|
|
|
assert(splice->next_[splice->height_] == nullptr);
|
|
|
|
for (int i = 0; i < splice->height_; ++i) {
|
|
|
|
assert(splice->next_[i] == nullptr ||
|
|
|
|
compare_(key, splice->next_[i]->Key()) < 0);
|
|
|
|
assert(splice->prev_[i] == head_ ||
|
|
|
|
compare_(splice->prev_[i]->Key(), key) <= 0);
|
|
|
|
assert(splice->prev_[i + 1] == splice->prev_[i] ||
|
|
|
|
splice->prev_[i + 1] == head_ ||
|
|
|
|
compare_(splice->prev_[i + 1]->Key(), splice->prev_[i]->Key()) <
|
|
|
|
0);
|
|
|
|
assert(splice->next_[i + 1] == splice->next_[i] ||
|
|
|
|
splice->next_[i + 1] == nullptr ||
|
|
|
|
compare_(splice->next_[i]->Key(), splice->next_[i + 1]->Key()) <
|
|
|
|
0);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
splice->height_ = 0;
|
|
|
|
}
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
bool InlineSkipList<Comparator>::Contains(const char* key) const {
|
|
|
|
Node* x = FindGreaterOrEqual(key);
|
InlineSkipList part 3/3 - new skiplist type that colocates key and node
Summary:
This diff completes the creation of InlineSkipList<Cmp>, which is like
SkipList<const char*, Cmp> but it always allocates the key contiguously
with the node. This allows us to remove the pointer from the node
to the key. As a result the memory usage of the skip list is reduced
(by 1 to sizeof(void*) bytes depending on the padding required to align
the key storage), cache locality is improved, and we halve the number
of calls to the allocator.
For skip lists whose keys are freshly-allocated const char*,
InlineSkipList is stricly preferrable to SkipList. This diff doesn't
replace SkipList, however, because some of the use cases of SkipList in
RocksDB are either character sequences that are not allocated at the
same time as the skip list node allocation (for example
hash_linklist_rep) or have different key types (for example
write_batch_with_index). Taking advantage of inline allocation for
those cases is left to future work.
The perf win is biggest for small values. For single-threaded CPU-bound
(32M fillrandom operations with no WAL log) with 16 byte keys and 0 byte
values, the db_bench perf goes from ~310k ops/sec to ~410k ops/sec. For
large values the improvement is less pronounced, but seems to be between
5% and 10% on the same configuration.
Test Plan: make check
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D51123
9 years ago
|
|
|
if (x != nullptr && Equal(key, x->Key())) {
|
|
|
|
return true;
|
|
|
|
} else {
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <class Comparator>
|
|
|
|
void InlineSkipList<Comparator>::TEST_Validate() const {
|
|
|
|
// Interate over all levels at the same time, and verify nodes appear in
|
|
|
|
// the right order, and nodes appear in upper level also appear in lower
|
|
|
|
// levels.
|
|
|
|
Node* nodes[kMaxPossibleHeight];
|
|
|
|
int max_height = GetMaxHeight();
|
|
|
|
assert(max_height > 0);
|
|
|
|
for (int i = 0; i < max_height; i++) {
|
|
|
|
nodes[i] = head_;
|
|
|
|
}
|
|
|
|
while (nodes[0] != nullptr) {
|
|
|
|
Node* l0_next = nodes[0]->Next(0);
|
|
|
|
if (l0_next == nullptr) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
assert(nodes[0] == head_ || compare_(nodes[0]->Key(), l0_next->Key()) < 0);
|
|
|
|
nodes[0] = l0_next;
|
|
|
|
|
|
|
|
int i = 1;
|
|
|
|
while (i < max_height) {
|
|
|
|
Node* next = nodes[i]->Next(i);
|
|
|
|
if (next == nullptr) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
auto cmp = compare_(nodes[0]->Key(), next->Key());
|
|
|
|
assert(cmp <= 0);
|
|
|
|
if (cmp == 0) {
|
|
|
|
assert(next == nodes[0]);
|
|
|
|
nodes[i] = next;
|
|
|
|
} else {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
i++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
for (int i = 1; i < max_height; i++) {
|
|
|
|
assert(nodes[i] != nullptr && nodes[i]->Next(i) == nullptr);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
} // namespace rocksdb
|