优化 Go 中的 map 并发存取

namemetricSource

map[string]*metricSourcenamemetricSource

简单实现如下：（为节省篇幅，省略了函数头和返回，只贴重要部分）

var source *memorySource
var present bool

p.lock.Lock() // lock the mutex
defer p.lock.Unlock() // unlock the mutex at the end

if source, present = p.sources[name]; !present {
	// The source wasn't found, so we'll create it.
	source = &memorySource{
		name: name,
		metrics: map[string]*memoryMetric{},
	}

	// Insert the newly created *memorySource.
	p.sources[name] = source
}

GOMAXPROCS

我们简化一下情况来说明这个问题，假设两个协程分别要获取“a”、“b”，并且“a”、“b”都已经存在于该 map 中。上述实现在运行时，一个协程获取到锁、拿指针、解锁、继续执行，此时另一个协程会被卡在获取锁。等待锁释放是非常耗时的，并且协程越多性能越差。

让它变快的方法之一是移除锁控制，并保证只有一个协程访问这个 map。这个方法虽然简单，但没有伸缩性。下面我们看看另一种简单的方法，并保证了线程安全和伸缩性。 

var source *memorySource
var present bool

if source, present = p.sources[name]; !present { // added this line
	// The source wasn't found, so we'll create it.

	p.lock.Lock() // lock the mutex
	defer p.lock.Unlock() // unlock at the end

	if source, present = p.sources[name]; !present {
		source = &memorySource{
			name: name,
			metrics: map[string]*memoryMetric{},
		}

		// Insert the newly created *memorySource.
		p.sources[name] = source
	}
	// if present is true, then another goroutine has already inserted
	// the element we want, and source is set to what we want.

} // added this line

// Note that if the source was present, we avoid the lock completely!

该实现可以达到 5,500,000 插入/秒，比第一个版本快 3.93 倍。有 4 个协程在跑测试，结果数值和预期是基本吻合的。

source

defer

var source *memorySource
var present bool

if source, present = p.sources[name]; !present {
	// The source wasn't found, so we'll create it.

	p.lock.Lock() // lock the mutex
	if source, present = p.sources[name]; !present {
		source = &memorySource{
			name: name,
			metrics: map[string]*memoryMetric{},
		}

		// Insert the newly created *memorySource.
		p.sources[name] = source
	}
	p.lock.Unlock() // unlock the mutex
}

// Note that if the source was present, we avoid the lock completely!

9,800,000 插入/秒！改了 4 行提升到 7 倍啊！！有木有！！！！

---

更新：（译注：原作者循序渐进非常赞）

上面实现正确么？No！通过 Go Data Race Detector 我们可以很轻松发现竟态条件，我们不能保证 map 在同时读写时的完整性。

RWMutex

var source *memorySource
var present bool

p.lock.RLock()
if source, present = p.sources[name]; !present {
	// The source wasn't found, so we'll create it.
	p.lock.RUnlock()
	p.lock.Lock()
	if source, present = p.sources[name]; !present {
		source = &memorySource{
			name: name,
			metrics: map[string]*memoryMetric{},
		}

		// Insert the newly created *memorySource.
		p.sources[name] = source
	}
	p.lock.Unlock()
} else {
	p.lock.RUnlock()
}

经测试，该版本性能为其之前版本的 93.8%，在保证正确性的前提先能到达这样已经很不错了。也许我们可以认为它们之间根本没有可比性，因为之前的版本是错的。

 

---- EOF ----