leveldb研究10- 流程分析：写数据

Posted on 2012-03-21 14:41 小明阅读(3509) 评论(0) 编辑收藏所属分类: 分布式计算

总体来说,leveldb的写操作有两个步骤，首先是针对log的append操作，然后是对memtable的插入操作。

影响写性能的因素有：
1. write_buffer_size
2. kL0_SlowdownWritesTrigger and kL0_StopWritesTrigger.提高这两个值，能够增加写的性能，但是降低读的性能

看看WriteOptions有哪些参数可以指定

struct WriteOptions {
  //设置sync=true,leveldb会调用fsync()，这会降低插入性能
  //同时会增加数据的安全性
  //Default: false
  bool sync;

  WriteOptions()
      : sync(false) {
  }
};

首先把Key,value转成WriteBatch

Status DB::Put(const WriteOptions& opt, const Slice& key, const Slice& value) {
  WriteBatch batch;
  batch.Put(key, value);
  return Write(opt, &batch);
}

接下来就是真正的插入了
这里使用了两把锁，主要是想提高并发能力，减少上锁的时间。
首先是检查是否可写，然后append log，最后是插入memtable
<db/dbimpl.cc>

Status DBImpl::Write(const WriteOptions& options, WriteBatch* updates) {
  Status status;
  //加锁
  MutexLock l(&mutex_);
  LoggerId self;
  //拿到写log的权利
  AcquireLoggingResponsibility(&self);
  //检查是否可写
  status = MakeRoomForWrite(false);  // May temporarily release lock and wait
  uint64_t last_sequence = versions_->LastSequence();
  if (status.ok()) {
    WriteBatchInternal::SetSequence(updates, last_sequence + 1);
    last_sequence += WriteBatchInternal::Count(updates);

    // Add to log and apply to memtable.  We can release the lock during
    // this phase since the "logger_" flag protects against concurrent
    // loggers and concurrent writes into mem_.
    {
      assert(logger_ == &self);
      mutex_.Unlock();
      //IO操作：写入LOG
      status = log_->AddRecord(WriteBatchInternal::Contents(updates));
      if (status.ok() && options.sync) {
        status = logfile_->Sync();
    &nbs