rocksdb/port/port_posix.cc
Nathan Bronson 7d87f02799 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
2015-12-25 11:03:40 -08:00

178 lines
4.7 KiB
C++

// Copyright (c) 2013, Facebook, Inc. All rights reserved.
// This source code is licensed under the BSD-style license found in the
// LICENSE file in the root directory of this source tree. An additional grant
// of patent rights can be found in the PATENTS file in the same directory.
//
// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file. See the AUTHORS file for names of contributors.
#include "port/port_posix.h"
#include <assert.h>
#if defined(__i386__) || defined(__x86_64__)
#include <cpuid.h>
#endif
#include <errno.h>
#include <signal.h>
#include <stdio.h>
#include <string.h>
#include <sys/time.h>
#include <sys/resource.h>
#include <unistd.h>
#include <cstdlib>
#include "util/logging.h"
namespace rocksdb {
namespace port {
static int PthreadCall(const char* label, int result) {
if (result != 0 && result != ETIMEDOUT) {
fprintf(stderr, "pthread %s: %s\n", label, strerror(result));
abort();
}
return result;
}
Mutex::Mutex(bool adaptive) {
#ifdef OS_LINUX
if (!adaptive) {
PthreadCall("init mutex", pthread_mutex_init(&mu_, nullptr));
} else {
pthread_mutexattr_t mutex_attr;
PthreadCall("init mutex attr", pthread_mutexattr_init(&mutex_attr));
PthreadCall("set mutex attr",
pthread_mutexattr_settype(&mutex_attr,
PTHREAD_MUTEX_ADAPTIVE_NP));
PthreadCall("init mutex", pthread_mutex_init(&mu_, &mutex_attr));
PthreadCall("destroy mutex attr",
pthread_mutexattr_destroy(&mutex_attr));
}
#else // ignore adaptive for non-linux platform
PthreadCall("init mutex", pthread_mutex_init(&mu_, nullptr));
#endif // OS_LINUX
}
Mutex::~Mutex() { PthreadCall("destroy mutex", pthread_mutex_destroy(&mu_)); }
void Mutex::Lock() {
PthreadCall("lock", pthread_mutex_lock(&mu_));
#ifndef NDEBUG
locked_ = true;
#endif
}
void Mutex::Unlock() {
#ifndef NDEBUG
locked_ = false;
#endif
PthreadCall("unlock", pthread_mutex_unlock(&mu_));
}
void Mutex::AssertHeld() {
#ifndef NDEBUG
assert(locked_);
#endif
}
CondVar::CondVar(Mutex* mu)
: mu_(mu) {
PthreadCall("init cv", pthread_cond_init(&cv_, nullptr));
}
CondVar::~CondVar() { PthreadCall("destroy cv", pthread_cond_destroy(&cv_)); }
void CondVar::Wait() {
#ifndef NDEBUG
mu_->locked_ = false;
#endif
PthreadCall("wait", pthread_cond_wait(&cv_, &mu_->mu_));
#ifndef NDEBUG
mu_->locked_ = true;
#endif
}
bool CondVar::TimedWait(uint64_t abs_time_us) {
struct timespec ts;
ts.tv_sec = static_cast<time_t>(abs_time_us / 1000000);
ts.tv_nsec = static_cast<suseconds_t>((abs_time_us % 1000000) * 1000);
#ifndef NDEBUG
mu_->locked_ = false;
#endif
int err = pthread_cond_timedwait(&cv_, &mu_->mu_, &ts);
#ifndef NDEBUG
mu_->locked_ = true;
#endif
if (err == ETIMEDOUT) {
return true;
}
if (err != 0) {
PthreadCall("timedwait", err);
}
return false;
}
void CondVar::Signal() {
PthreadCall("signal", pthread_cond_signal(&cv_));
}
void CondVar::SignalAll() {
PthreadCall("broadcast", pthread_cond_broadcast(&cv_));
}
RWMutex::RWMutex() {
PthreadCall("init mutex", pthread_rwlock_init(&mu_, nullptr));
}
RWMutex::~RWMutex() { PthreadCall("destroy mutex", pthread_rwlock_destroy(&mu_)); }
void RWMutex::ReadLock() { PthreadCall("read lock", pthread_rwlock_rdlock(&mu_)); }
void RWMutex::WriteLock() { PthreadCall("write lock", pthread_rwlock_wrlock(&mu_)); }
void RWMutex::ReadUnlock() { PthreadCall("read unlock", pthread_rwlock_unlock(&mu_)); }
void RWMutex::WriteUnlock() { PthreadCall("write unlock", pthread_rwlock_unlock(&mu_)); }
int PhysicalCoreID() {
#if defined(__i386__) || defined(__x86_64__)
// if you ever find that this function is hot on Linux, you can go from
// ~200 nanos to ~20 nanos by adding the machinery to use __vdso_getcpu
unsigned eax, ebx = 0, ecx, edx;
__get_cpuid(1, &eax, &ebx, &ecx, &edx);
return ebx >> 24;
#else
// getcpu or sched_getcpu could work here
return -1;
#endif
}
void InitOnce(OnceType* once, void (*initializer)()) {
PthreadCall("once", pthread_once(once, initializer));
}
void Crash(const std::string& srcfile, int srcline) {
fprintf(stdout, "Crashing at %s:%d\n", srcfile.c_str(), srcline);
fflush(stdout);
kill(getpid(), SIGTERM);
}
int GetMaxOpenFiles() {
#if defined(RLIMIT_NOFILE)
struct rlimit no_files_limit;
if (getrlimit(RLIMIT_NOFILE, &no_files_limit) != 0) {
return -1;
}
// protect against overflow
if (no_files_limit.rlim_cur >= std::numeric_limits<int>::max()) {
return std::numeric_limits<int>::max();
}
return static_cast<int>(no_files_limit.rlim_cur);
#endif
return -1;
}
} // namespace port
} // namespace rocksdb