go-zero 给咱们提供了两种限流器,而且都是基于 redis 实现的可分布式的
限流器外围文件带正文代码如下,大家能够参阅
- 计数器限流器 https://github.com/TTSimple/g...
- 令牌桶限流器 https://github.com/TTSimple/g...
咱们通过最小化代码来看看限流器的外围思路
繁难计数器算法
// 繁难计数器算法type Counter struct { rate int // 计数周期内最多容许的申请数 begin time.Time // 计数开始工夫 cycle time.Duration // 计数周期 count int // 计数周期内累计收到的申请数 lock sync.Mutex}func (l *Counter) Allow() bool { l.lock.Lock() defer l.lock.Unlock() if l.count == l.rate-1 { now := time.Now() if now.Sub(l.begin) >= l.cycle { // 速度容许范畴内, 重置计数器 l.Reset(now) return true } else { return false } } else { // 没有达到速率限度,计数加1 l.count++ return true }}func (l *Counter) Set(r int, cycle time.Duration) { l.rate = r l.begin = time.Now() l.cycle = cycle l.count = 0}func (l *Counter) Reset(t time.Time) { l.begin = t l.count = 0}func Test_Counter(t *testing.T) { c := Counter{} c.Set(20, time.Second) reqTime := 2 * time.Second // 总申请工夫 reqNum := 200 // 总申请次数 reqInterval := reqTime / time.Duration(reqNum) // 每次申请距离 var trueCount, falseCount int for i := 0; i < reqNum; i++ { go func() { if c.Allow() { trueCount++ } else { falseCount++ } }() time.Sleep(reqInterval) } fmt.Println("true count: ", trueCount) fmt.Println("false count: ", falseCount)}
最终输入
// === RUN Test_Counter// true count: 44// false count: 156// --- PASS: Test_Counter (2.07s)
繁难令牌桶算法
// 繁难令牌桶算法type TokenBucket struct { rate int64 // 固定的token放入速率, r/s capacity int64 // 桶的容量 tokens int64 // 桶中以后token数量 lastTokenSec int64 // 桶上次放token的工夫戳 s lock sync.Mutex}// 判断是否可通过func (l *TokenBucket) Allow() bool { l.lock.Lock() defer l.lock.Unlock() now := time.Now().Unix() // 先增加初始令牌 l.tokens = l.tokens + (now-l.lastTokenSec)*l.rate if l.tokens > l.capacity { l.tokens = l.capacity } l.lastTokenSec = now if l.tokens > 0 { // 还有令牌,支付令牌 l.tokens-- return true } // 没有令牌,则回绝 return false}// 动静设置参数// r rate// c capacityfunc (l *TokenBucket) Set(r, c int64) { l.rate = r l.capacity = c l.tokens = r l.lastTokenSec = time.Now().Unix()}func Test_TokenBucket(t *testing.T) { lb := &TokenBucket{} lb.Set(20, 20) requestTime := 2 * time.Second // 总申请工夫 requestNum := 200 // 总申请次数 requestInterval := requestTime / time.Duration(requestNum) // 每次申请距离 var trueCount, falseCount int for i := 0; i < requestNum; i++ { go func() { if lb.Allow() { trueCount++ } else { falseCount++ } }() time.Sleep(requestInterval) } fmt.Println("true count: ", trueCount) fmt.Println("false count: ", falseCount)}
最终输入
=== RUN Test_TokenBuckettrue count: 60false count: 140--- PASS: Test_TokenBucket (2.07s)
繁难漏桶算法
漏桶算法的分布式版本 go-zero 没有给咱们实现,咱们看看其外围算法,而后参照外围算法来实现分布式版本,给大家安排个作业 :)
// 繁难漏桶算法type LeakyBucket struct { rate float64 // 固定每秒出水速率 capacity float64 // 桶的容量 water float64 // 桶中以后水量 lastLeakMs int64 // 桶上次漏水工夫戳 ms lock sync.Mutex}// 判断是否可通过func (l *LeakyBucket) Allow() bool { l.lock.Lock() defer l.lock.Unlock() now := time.Now().UnixNano() / 1e6 eclipse := float64((now - l.lastLeakMs)) * l.rate / 1000 // 先执行漏水 l.water = l.water - eclipse // 计算残余水量 l.water = math.Max(0, l.water) // 桶干了 l.lastLeakMs = now if (l.water + 1) < l.capacity { // 尝试加水,并且水还未满 l.water++ return true } else { // 水满,回绝加水 return false }}// 动静设置参数// r rate// c capacityfunc (l *LeakyBucket) Set(r, c float64) { l.rate = r l.capacity = c l.water = 0 l.lastLeakMs = time.Now().UnixNano() / 1e6}func Test_LeakyBucket(t *testing.T) { lb := &LeakyBucket{} lb.Set(20, 20) reqTime := 2 * time.Second // 总申请工夫 reqNum := 200 // 总申请次数 reqInterval := reqTime / time.Duration(reqNum) // 每次申请距离 var trueCount, falseCount int for i := 0; i < reqNum; i++ { go func() { if lb.Allow() { trueCount++ } else { falseCount++ } }() time.Sleep(reqInterval) } fmt.Println("true count: ", trueCount) fmt.Println("false count: ", falseCount)}
最终输入
// === RUN Test_LeakyBucket// true count: 60// false count: 140// --- PASS: Test_LeakyBucket (2.06s)