ZenQ

> DevOps

开源

一个线程安全的队列,比golang的原生通道更快速、更节省资源

695 stars0 点赞0 次浏览

访问官网 GitHub

工具介绍

一个线程安全的队列,比golang的原生通道更快速、更节省资源

ZenQ

A low-latency thread-safe queue in golang implemented using a lock-free ringbuffer and runtime internals

Based on the LMAX Disruptor Pattern

Features

Much faster than native channels in both SPSC (single-producer-single-consumer) and MPSC (multi-producer-single-consumer) modes in terms of time/op
More resource efficient in terms of memory_allocation/op and num_allocations/op evident while benchmarking large batch size inputs
Handles the case where NUM_WRITER_GOROUTINES > NUM_CPU_CORES much better than native channels
Selection from multiple ZenQs just like golang's select{} ensuring fair selection and no starvation
Closing a ZenQ

Benchmarks to support the above claims here

Installation

You need Golang 1.19.x or above

bash

$ go get github.com/alphadose/zenq/v2

Usage

Simple Read/Write

package main

import (
	"fmt"

	"github.com/alphadose/zenq/v2"
)

type payload struct {
	alpha int
	beta  string
}

func main() {
	zq := zenq.New[payload](10)

	for j := 0; j  The number of goroutines concurrently writing to ZenQ/Channel
* INPUT_SIZE -> The number of input payloads to be passed through ZenQ/Channel from producers to consumer

```bash
Computed from benchstat of 30 benchmarks each via go test -benchmem -bench=. benchmarks/simple/*.go

name                                     time/op
_Chan_NumWriters1_InputSize600-8          23.2µs ± 1%
_ZenQ_NumWriters1_InputSize600-8          17.9µs ± 1%
_Chan_NumWriters3_InputSize60000-8        5.27ms ± 3%
_ZenQ_NumWriters3_InputSize60000-8        2.36ms ± 2%
_Chan_NumWriters8_InputSize6000000-8       671ms ± 2%
_ZenQ_NumWriters8_InputSize6000000-8       234ms ± 6%
_Chan_NumWriters100_InputSize6000000-8     1.59s ± 4%
_ZenQ_NumWriters100_InputSize6000000-8     309ms ± 2%
_Chan_NumWriters1000_InputSize7000000-8    1.97s ± 0%
_ZenQ_NumWriters1000_InputSize7000000-8    389ms ± 4%
_Chan_Million_Blocking_Writers-8           10.4s ± 2%
_ZenQ_Million_Blocking_Writers-8           2.32s ±21%

name                                     alloc/op
_Chan_NumWriters1_InputSize600-8           0.00B
_ZenQ_NumWriters1_InputSize600-8           0.00B
_Chan_NumWriters3_InputSize60000-8          109B ±68%
_ZenQ_NumWriters3_InputSize60000-8        24.6B ±107%
_Chan_NumWriters8_InputSize6000000-8       802B ±241%
_ZenQ_NumWriters8_InputSize6000000-8     1.18kB ±100%
_Chan_NumWriters100_InputSize6000000-8    44.2kB ±41%
_ZenQ_NumWriters100_InputSize6000000-8    10.7kB ±38%
_Chan_NumWriters1000_InputSize7000000-8    476kB ± 8%
_ZenQ_NumWriters1000_InputSize7000000-8   90.6kB ±10%
_Chan_Million_Blocking_Writers-8           553MB ± 0%
_ZenQ_Million_Blocking_Writers-8           122MB ± 3%

name                                     allocs/op
_Chan_NumWriters1_InputSize600-8            0.00
_ZenQ_NumWriters1_InputSize600-8            0.00
_Chan_NumWriters3_InputSize60000-8          0.00
_ZenQ_NumWriters3_InputSize60000-8          0.00
_Chan_NumWriters8_InputSize6000000-8       2.76 ±190%
_ZenQ_NumWriters8_InputSize6000000-8        5.47 ±83%
_Chan_NumWriters100_InputSize6000000-8       159 ±26%
_ZenQ_NumWriters100_InputSize6000000-8      25.1 ±39%
_Chan_NumWriters1000_InputSize7000000-8    1.76k ± 6%
_ZenQ_NumWriters1000_InputSize7000000-8     47.3 ±31%
_Chan_Million_Blocking_Writers-8           2.00M ± 0%
_ZenQ_Million_Blocking_Writers-8           1.00M ± 0%

The above results show that ZenQ is more efficient than channels in all 3 metrics i.e time/op, mem_alloc/op and num_allocs/op for the following tested cases:-

SPSC
MPSC with NUM_WRITER_GOROUTINES NUM_CPU_CORES

Cherry on the Cake

In SPSC mode ZenQ is faster than channels by 92 seconds in case of input size of 6 * 108 elements

bash

❯ go run benchmarks/simple/main.go

With Input Batch Size: 60 and Num Concurrent Writers: 1

Native Channel Runner completed transfer in: 26.916µs
ZenQ Runner completed transfer in: 20.292µs
====================================================================

With Input Batch Size: 600 and Num Concurrent Writers: 1

Native Channel Runner completed transfer in: 135.75µs
ZenQ Runner completed transfer in: 105.792µs
====================================================================

With Input Batch Size: 6000 and Num Concurrent Writers: 1

Native Channel Runner completed transfer in: 2.100209ms
ZenQ Runner completed transfer in: 510.792µs
====================================================================

With Input Batch Size: 6000000 and Num Concurrent Writers: 1

Native Channel Runner completed transfer in: 1.241481917s
ZenQ Runner completed transfer in: 226.068209ms
====================================================================

With Input Batch Size: 600000000 and Num Concurrent Writers: 1

Native Channel Runner completed transfer in: 1m55.074638875s
ZenQ Runner completed transfer in: 22.582667917s
====================================================================