Nginx下Go的多种使用方式性能比较

时间:2022-10-27 19:25:25

结论:Go HTTP standalone >(优于)   Nginx proxy to Go HTTP >(优于) Nginx fastcgi to Go TCP FastCGI


原文链接:http://www.oschina.net/translate/benchmarking-nginx-with-go?from=20131222

英文链接:https://gist.github.com/hgfischer/7965620

目前有很多提供Go语言HTTP应用服务的方法,但其中最好的选择取决于每个应用的实际情况。目前,Nginx看起来是每个新项目的标准Web服务器,即使在有其他许多不错Web服务器的情况下。然而,在Nginx上提供Go应用服务的开销是多少呢?我们需要一些nginx的特性参数(vhosts,负载均衡,缓存,等等)或者直接使用Go提供服务?如果你需要nginx,最快的连接机制是什么?这就是在这我试图回答的问题。该基准测试的目的不是要验证Go比nginx的快或慢。那将会很愚蠢。

下面是我们要比较不同的设置:

  • Go HTTP standalone (as the control group)
  • Nginx proxy to Go HTTP
  • Nginx fastcgi to Go TCP FastCGI
  • Nginx fastcgi to Go Unix Socket FastCGI

硬件

因为我们将在相同的硬件下比较所有设置,硬件选择的是廉价的一个。这不应该是一个大问题。

  • Samsung 笔记本 NP550P5C-AD1BR
  • Intel Core i7 3630QM @2.4GHz (quad core, 8 threads)
  • CPU caches: (L1: 256KiB, L2: 1MiB, L3: 6MiB)
  • RAM 8GiB DDR3 1600MHz

软件

  • Ubuntu 13.10 amd64 Saucy Salamander (updated)
  • Nginx 1.4.4 (1.4.4-1~saucy0 amd64)
  • Go 1.2 (linux/amd64)
  • wrk 3.0.4

设置

内核

只需很小的一点调整,将内核的limits调高。如果你对这一变量有更好的想法,请在写在下面评论处:

01 fs.file-max                    9999999
02 fs.nr_open                     9999999
03 net.core.netdev_max_backlog    4096
04 net.core.rmem_max              16777216
05 net.core.somaxconn             65535
06 net.core.wmem_max              16777216
07 net.ipv4.ip_forward            0
08 net.ipv4.ip_local_port_range   1025       65535
09 net.ipv4.tcp_fin_timeout       30
10 net.ipv4.tcp_keepalive_time    30
11 net.ipv4.tcp_max_syn_backlog   20480
12 net.ipv4.tcp_max_tw_buckets    400000
13 net.ipv4.tcp_no_metrics_save   1
14 net.ipv4.tcp_syn_retries       2
15 net.ipv4.tcp_synack_retries    2
16 net.ipv4.tcp_tw_recycle        1
17 net.ipv4.tcp_tw_reuse          1
18 vm.min_free_kbytes             65536
19 vm.overcommit_memory           1

Limits

供root和www-data打开的最大文件数限制被配置为200000。

Nginx

有几个必需得Nginx调整。有人跟我说过,我禁用了gzip以保证比较公平。下面是它的配置文件/etc/nginx/nginx.conf:

01 user www-data;
02 worker_processes auto;
03 worker_rlimit_nofile 200000;
04 pid /var/run/nginx.pid;
05  
06 events {
07     worker_connections 10000;
08     use epoll;
09     multi_accept on;
10 }
11  
12 http {
13     sendfile on;
14     tcp_nopush on;
15     tcp_nodelay on;
16     keepalive_timeout 300;
17     keepalive_requests 10000;
18     types_hash_max_size 2048;
19  
20     open_file_cache max=200000 inactive=300s;
21     open_file_cache_valid 300s;
22     open_file_cache_min_uses 2;
23     open_file_cache_errors on;
24  
25     server_tokens off;
26     dav_methods off;
27  
28     include /etc/nginx/mime.types;
29     default_type application/octet-stream;
30  
31     access_log /var/log/nginx/access.log combined;
32     error_log /var/log/nginx/error.log warn;
33  
34     gzip off;
35     gzip_vary off;
36  
37     include /etc/nginx/conf.d/*.conf;
38     include /etc/nginx/sites-enabled/*.conf;
39 }

Nginx vhosts

01 upstream go_http {
02     server 127.0.0.1:8080;
03     keepalive 300;
04 }
05  
06 server {
07     listen 80;
08     server_name go.http;
09     access_log off;
10     error_log /dev/null crit;
11  
12     location / {
13         proxy_pass http://go_http;
14         proxy_http_version 1.1;
15         proxy_set_header Connection "";
16     }
17 }
18  
19 upstream go_fcgi_tcp {
20     server 127.0.0.1:9001;
21     keepalive 300;
22 }
23  
24 server {
25     listen 80;
26     server_name go.fcgi.tcp;
27     access_log off;
28     error_log /dev/null crit;
29  
30     location / {
31         include fastcgi_params;
32         fastcgi_keep_conn on;
33         fastcgi_pass go_fcgi_tcp;
34     }
35 }
36  
37 upstream go_fcgi_unix {
38     server unix:/tmp/go.sock;
39     keepalive 300;
40 }
41  
42 server {
43     listen 80;
44     server_name go.fcgi.unix;
45     access_log off;
46     error_log /dev/null crit;
47  
48     location / {
49         include fastcgi_params;
50         fastcgi_keep_conn on;
51         fastcgi_pass go_fcgi_unix;
52     }
53 }

Go源码

01 package main
02  
03 import (
04     "fmt"
05     "log"
06     "net"
07     "net/http"
08     "net/http/fcgi"
09     "os"
10     "os/signal"
11     "syscall"
12 )
13  
14 var (
15     abort bool
16 )
17  
18 const (
19     SOCK = "/tmp/go.sock"
20 )
21  
22 type Server struct {
23 }
24  
25 func (s Server) ServeHTTP(w http.ResponseWriter, r *http.Request) {
26     body := "Hello World\n"
27     // Try to keep the same amount of headers
28     w.Header().Set("Server""gophr")
29     w.Header().Set("Connection""keep-alive")
30     w.Header().Set("Content-Type""text/plain")
31     w.Header().Set("Content-Length", fmt.Sprint(len(body)))
32     fmt.Fprint(w, body)
33 }
34  
35 func main() {
36     sigchan := make(chan os.Signal, 1)
37     signal.Notify(sigchan, os.Interrupt)
38     signal.Notify(sigchan, syscall.SIGTERM)
39  
40     server := Server{}
41  
42     go func() {
43         http.Handle("/", server)
44         if err := http.ListenAndServe(":8080", nil); err != nil {
45             log.Fatal(err)
46         }
47     }()
48  
49     go func() {
50         tcp, err := net.Listen("tcp"":9001")
51         if err != nil {
52             log.Fatal(err)
53         }
54         fcgi.Serve(tcp, server)
55     }()
56  
57     go func() {
58         unix, err := net.Listen("unix", SOCK)
59         if err != nil {
60             log.Fatal(err)
61         }
62         fcgi.Serve(unix, server)
63     }()
64  
65     <-sigchan
66  
67     if err := os.Remove(SOCK); err != nil {
68         log.Fatal(err)
69     }
70 }

检查HTTP header

为公平起见,所有的请求必需大小相同。

01 $ curl -sI http://127.0.0.1:8080/
02 HTTP/1.1 200 OK
03 Connection: keep-alive
04 Content-Length: 12
05 Content-Type: text/plain
06 Server: gophr
07 Date: Sun, 15 Dec 2013 14:59:14 GMT
08  
09 $ curl -sI http://127.0.0.1:8080/ | wc -c
10 141
01 $ curl -sI http://go.http/
02 HTTP/1.1 200 OK
03 Server: nginx
04 Date: Sun, 15 Dec 2013 14:59:31 GMT
05 Content-Type: text/plain
06 Content-Length: 12
07 Connection: keep-alive
08  
09 $ curl -sI http://go.http/ | wc -c
10 141
01 $ curl -sI http://go.fcgi.tcp/
02 HTTP/1.1 200 OK
03 Content-Type: text/plain
04 Content-Length: 12
05 Connection: keep-alive
06 Date: Sun, 15 Dec 2013 14:59:40 GMT
07 Server: gophr
08  
09 $ curl -sI http://go.fcgi.tcp/ | wc -c
10 141
01 $ curl -sI http://go.fcgi.unix/
02 HTTP/1.1 200 OK
03 Content-Type: text/plain
04 Content-Length: 12
05 Connection: keep-alive
06 Date: Sun, 15 Dec 2013 15:00:15 GMT
07 Server: gophr
08  
09 $ curl -sI http://go.fcgi.unix/ | wc -c
10 141

启动引擎

  • 使用sysctl配置内核
  • 配置Nginx
  • 配置Nginx vhosts
  • 用www-data启动服务
  • 运行基准测试

基准测试

GOMAXPROCS = 1

Go standalone
1 # wrk -t100 -c5000 -d30s http://127.0.0.1:8080/
2 Running 30s test http://127.0.0.1:8080/
3   100 threads and 5000 connections
4   Thread Stats   Avg      Stdev     Max   +/- Stdev
5     Latency   116.96ms   17.76ms 173.96ms   85.31%
6     Req/Sec   429.16     49.20   589.00     69.44%
7   1281567 requests in 29.98s, 215.11MB read
8 Requests/sec:  42745.15
9 Transfer/sec:      7.17MB
Nginx + Go through HTTP
1 # wrk -t100 -c5000 -d30s http://go.http/
2 Running 30s test http://go.http/
3   100 threads and 5000 connections
4   Thread Stats   Avg      Stdev     Max   +/- Stdev
5     Latency   124.57ms   18.26ms 209.70ms   80.17%
6     Req/Sec   406.29     56.94     0.87k    89.41%
7   1198450 requests in 29.97s, 201.16MB read
8 Requests/sec:  39991.57
9 Transfer/sec:      6.71MB
Nginx + Go through FastCGI TCP
01 # wrk -t100 -c5000 -d30s http://go.fcgi.tcp/
02 Running 30s test http://go.fcgi.tcp/
03   100 threads and 5000 connections
04   Thread Stats   Avg      Stdev     Max   +/- Stdev
05     Latency   514.57ms  119.80ms   1.21s    71.85%
06     Req/Sec    97.18     22.56   263.00     79.59%
07   287416 requests in 30.00s, 48.24MB read
08   Socket errors: connect 0, read 0, write 0, timeout 661
09 Requests/sec:   9580.75
10 Transfer/sec:      1.61MB
Nginx + Go through FastCGI Unix Socket
01 # wrk -t100 -c5000 -d30s http://go.fcgi.unix/
02 Running 30s test http://go.fcgi.unix/
03   100 threads and 5000 connections
04   Thread Stats   Avg      Stdev     Max   +/- Stdev
05     Latency   425.64ms   80.53ms 925.03ms   76.88%
06     Req/Sec   117.03     22.13   255.00     81.30%
07   350162 requests in 30.00s, 58.77MB read
08   Socket errors: connect 0, read 0, write 0, timeout 210
09 Requests/sec:  11670.72
10 Transfer/sec:      1.96MB

GOMAXPROCS = 8

Go standalone
1 # wrk -t100 -c5000 -d30s http://127.0.0.1:8080/
2 Running 30s test http://127.0.0.1:8080/
3   100 threads and 5000 connections
4   Thread Stats   Avg      Stdev     Max   +/- Stdev
5     Latency    39.25ms    8.49ms  86.45ms   81.39%
6     Req/Sec     1.29k   129.27     1.79k    69.23%
7   3837995 requests in 29.89s, 644.19MB read
8 Requests/sec: 128402.88
9 Transfer/sec:     21.55MB
Nginx + Go through HTTP
1 # wrk -t100 -c5000 -d30s http://go.http/
2 Running 30s test http://go.http/
3   100 threads and 5000 connections
4   Thread Stats   Avg      Stdev     Max   +/- Stdev
5     Latency   336.77ms  297.88ms 632.52ms   60.16%
6     Req/Sec     2.36k     2.99k   19.11k    84.83%
7   2232068 requests in 29.98s, 374.64MB read
8 Requests/sec:  74442.91
9 Transfer/sec:     12.49MB
Nginx + Go through FastCGI TCP
01 # wrk -t100 -c5000 -d30s http://go.fcgi.tcp/
02 Running 30s test http://go.fcgi.tcp/
03   100 threads and 5000 connections
04   Thread Stats   Avg      Stdev     Max   +/- Stdev
05     Latency   217.69ms  121.22ms   1.80s    75.14%
06     Req/Sec   263.09    102.78   629.00     62.54%
07   721027 requests in 30.01s, 121.02MB read
08   Socket errors: connect 0, read 0, write 176, timeout 1343
09 Requests/sec:  24026.50
10 Transfer/sec:      4.03MB
Nginx + Go through FastCGI Unix Socket
1 # wrk -t100 -c5000 -d30s http://go.fcgi.unix/
2 Running 30s test http://go.fcgi.unix/
3   100 threads and 5000 connections
4   Thread Stats   Avg      Stdev     Max   +/- Stdev
5     Latency   694.32ms  332.27ms   1.79s    62.13%
6     Req/Sec   646.86    669.65     6.11k    87.80%
7   909836 requests in 30.00s, 152.71MB read
8 Requests/sec:  30324.77
9 Transfer/sec:      5.09MB

结论

第一组基准测试时一些Nginx的设置还没有很好的优化(启用gzip,Go的后端没有使用keep-alive连接)。当改为wrk以及按建议优化Nginx后结果有较大差异。

当GOMAXPROCS=1时,Nginx的开销不是那么大,但当OMAXPROCS=8时差异就很大了。以后可能会再试一下其他设置。如果你需要使用Nginx像虚拟主机,负载均衡,缓存等特性,使用HTTP proxy,别使用FastCGI。有些人说Go的FastCGI还没有被很好优化,这也许就是测试结果中巨大差异的原因。