@duanyubin 2016-04-12T03:06:04.000000Z 字数 4168 阅读 455

通过nodejs 学习后端

NODE

谈谈后端

提供http接口
存储数据
与语言无关

Node.js

Node.js uses an event-driven, non-blocking I/O model that makes it lightweight and efficient.

Node.js 基础架构

事件驱动与事件循环

事件驱动类似于浏览器中的鼠标事件键盘事件等
Node.js 程序由事件循环开始，到事件循环结束，所有的逻辑都是事件的回调函数，所以 Node.js始终在事件循环中，程序入口就是事件循环第一个事件的回调函数

非阻塞I/O

不同介质下的I/O操作耗时

I/O	CPU Cycle
L1-cache	3
L2-cache	14
RAM	250
Disk	41000000
Network	240000000

同步阻塞式I/O

阻塞I/O

非阻塞I/O

此处输入图片的描述

优点

减少了多线程的开销
避免线程创建和调度
充分利用cpu
避免了锁的问题
更好的支持高并发

解决并发常见的模型

单进程多线程

每个 request 都在单独的线程中处理，即使某一个请求发生很严重的错误也不会影响到其它请求

单线程多进程

通过fork的方法，依据CPU核心数，创建对应的进程数

nodejs 解决单进程问题

1. 多进程监听同一端口

多个进程之间会竞争 accpet 一个连接，产生惊群现象，效率比较低。
由于无法控制一个新的连接由哪个进程来处理，必然导致各 worker 进程之间的负载非常不均衡。

2. 多进程监听不同端口

通过nginx处理负载均衡

http { 
  upstream cluster { 
      server 127.0.0.1:3000; 
      server 127.0.0.1:3001; 
      server 127.0.0.1:3002; 
      server 127.0.0.1:3003; 
  } 
  server { 
       listen 80; 
       server_name www.domain.com; 
       location / { 
            proxy_pass http://cluster;
       } 
  }
}

3. cluster

var cluster = require('cluster');  
var http    = require('http');  
var os      = require('os');
var numCPUs = os.cpus().length;
if (cluster.isMaster) {  
  // Master:
  // Let's fork as many workers as you have CPU cores
  for (var i = 0; i < numCPUs; ++i) {
    cluster.fork();
  }
} else {
  // Worker:
  // Let's spawn a HTTP server
  // (Workers can share any TCP connection.
  //  In this case its a HTTP server)
  http.createServer(function(req, res) {
    res.writeHead(200);
    res.end("hello world");
  }).listen(8080);
}

4. pm2

pm2 start app.js -i 0

性能测试

测试代码

const http = require('http')
const hostname = '127.0.0.1'
const port = 3000
const server = http.createServer((req, res) => {
  res.statusCode = 200
  res.setHeader('Content-Type', 'text/plain')
  res.end('Hello World\n')
})
server.listen(port, hostname, () => {
  console.log(`Server running at http://${hostname}:${port}/`)
})

测试工具

naughtyDog:~ dyb$ loadtest -n 10000 -c 100 http://127.0.0.1:3000/

单进程

naughtyDog:~ dyb$ loadtest -n 10000 -c 100 http://127.0.0.1:3000/
[Mon Apr 11 2016 14:11:43 GMT+0800 (CST)] INFO Requests: 0 (0%), requests per second: 0, mean latency: 0 ms
[Mon Apr 11 2016 14:11:48 GMT+0800 (CST)] INFO Requests: 6378 (64%), requests per second: 1275, mean latency: 60 ms
[Mon Apr 11 2016 14:11:53 GMT+0800 (CST)] INFO Requests: 8148 (81%), requests per second: 353, mean latency: 340 ms
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO 
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO Target URL:          http://127.0.0.1:3000/
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO Max requests:        10000
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO Concurrency level:   100
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO Agent:               none
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO 
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO Completed requests:  10000
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO Total errors:        0
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO Total time:          12.325297339 s
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO Requests per second: 811
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO Total time:          12.325297339 s
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO 
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO Percentage of the requests served within a certain time
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO   50%      60 ms
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO   90%      83 ms
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO   95%      98 ms
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO   99%      4052 ms
[Mon Apr 11 2016 14:11:55 GMT+0800 (CST)] INFO  100%      4103 ms (longest request)

多进程

naughtyDog:~ dyb$ loadtest -n 10000 -c 100 http://127.0.0.1:3000/
[Mon Apr 11 2016 14:11:19 GMT+0800 (CST)] INFO Requests: 0 (0%), requests per second: 0, mean latency: 0 ms
[Mon Apr 11 2016 14:11:24 GMT+0800 (CST)] INFO Requests: 7812 (78%), requests per second: 1560, mean latency: 60 ms
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO 
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO Target URL:          http://127.0.0.1:3000/
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO Max requests:        10000
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO Concurrency level:   100
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO Agent:               none
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO 
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO Completed requests:  10000
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO Total errors:        0
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO Total time:          7.234391098 s
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO Requests per second: 1382
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO Total time:          7.234391098 s
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO 
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO Percentage of the requests served within a certain time
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO   50%      62 ms
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO   90%      85 ms
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO   95%      96 ms
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO   99%      624 ms
[Mon Apr 11 2016 14:11:26 GMT+0800 (CST)] INFO  100%      627 ms (longest request)

Node.js不适用场景

计算密集型程序
单用户多任务应用
逻辑十分复杂的事务

通过nodejs 学习后端

谈谈后端

Node.js

事件驱动与事件循环

非阻塞I/O

不同介质下的I/O操作耗时

同步阻塞式I/O

非阻塞I/O

优点

解决并发常见的模型

单进程多线程

单线程多进程

nodejs 解决单进程问题

1. 多进程监听同一端口

2. 多进程监听不同端口

3. cluster

4. pm2

性能测试

测试代码

测试工具

单进程

多进程

Node.js不适用场景

内容目录