我们知道压力测试的软件确实很多，诸如微软的WAST，惠普的LoadRunner以及等等其他的，但这些软件学习起来还是需要花费些时间，在选择上实在头痛，后来在郭欣的那本《构建高性能WEB站点》上看到了他介绍的这款Apache自带的压力测试工具ab，十分喜爱，于是今天终于有机会体验下ab对网站的压力测试。
实验之前我的apache已经安装了，操作系统：Ubuntu 10.04 VMware 7.0
1、先查看一下版本信息 ab -V（注意是大写的V）
- studiogang@studiogang:~$ ab -V
- This is ApacheBench, Version 2.3 <$Revision: 655654 $>
- Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
- Licensed to The Apache Software Foundation, http://www.apache.org/
2、我们也可以使用小写的v查看下ab命令的一些属性 ab -v
- studiogang@studiogang:~$ ab -v
- ab: option requires an argument -- v
- ab: wrong number of arguments
- Usage: ab [options] [http[s]://]hostname[:port]/path
- Options are:
- -n requests Number of requests to perform
- -c concurrency Number of multiple requests to make
- -t timelimit Seconds to max. wait for responses
- -b windowsize Size of TCP send/receive buffer, in bytes
- -p postfile File containing data to POST. Remember also to set -T
- -u putfile File containing data to PUT. Remember also to set -T
- -T content-type Content-type header for POSTing, eg.
- 'application/x-www-form-urlencoded'
- Default is 'text/plain'
- -v verbosity How much troubleshooting info to print
- -w Print out results in HTML tables
- -i Use HEAD instead of GET
- -x attributes String to insert as table attributes
- -y attributes String to insert as tr attributes
- -z attributes String to insert as td or th attributes
- -C attribute Add cookie, eg. 'Apache=1234. (repeatable)
- -H attribute Add Arbitrary header line, eg. 'Accept-Encoding: gzip'
- Inserted after all normal header lines. (repeatable)
- -A attribute Add Basic WWW Authentication, the attributes
- are a colon separated username and password.
- -P attribute Add Basic Proxy Authentication, the attributes
- are a colon separated username and password.
- -X proxy:port Proxyserver and port number to use
- -V Print version number and exit
- -k Use HTTP KeepAlive feature
- -d Do not show percentiles served table.
- -S Do not show confidence estimators and warnings.
- -g filename Output collected data to gnuplot format file.
- -e filename Output CSV file with percentages served
- -r Don't exit on socket receive errors.
- -h Display usage information (this message)
- -Z ciphersuite Specify SSL/TLS cipher suite (See openssl ciphers)
- -f protocol Specify SSL/TLS protocol (SSL2, SSL3, TLS1, or ALL)
3、现在我们就对51CTO的网站进行一次压力测试吧，使用命令ab -n1000 -c10 http://www.51cto.com/index.php，其中 -n1000 表示总请求数 -c10表示并发用户数为10 http://www.51cto.com/index.php 表示请求的URL，下面是测试的结果，其中我们最关心的三个指标，我已经注释出来了。
- studiogang@studiogang:~$ ab -n1000 -c10 http://www.51cto.com/index.php
- This is ApacheBench, Version 2.3 <$Revision: 655654 $>
- Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
- Licensed to The Apache Software Foundation, http://www.apache.org/
-
-
- Benchmarking www.51cto.com (be patient)
- Completed 100 requests
- Completed 200 requests
- Completed 300 requests
- Completed 400 requests
- Completed 500 requests
- Completed 600 requests
- Completed 700 requests
- Completed 800 requests
- Completed 900 requests
- Completed 1000 requests
- Finished 1000 requests
-
- /*WEB服务器用的是nginx*/
- Server Software: nginx
- Server Hostname: www.51cto.com
- Server Port: 80
-
- Document Path: /index.php
- Document Length: 154 bytes
-
- Concurrency Level: 10
- Time taken for tests: 74.373 seconds
- Complete requests: 1000
- Failed requests: 0
- Write errors: 0
- Non-2xx responses: 1000
- Total transferred: 330000 bytes
- HTML transferred: 154000 bytes
- /*大家最关心的指标之一，指的是吞吐率
- 相当于 LR 中的 每秒事务数 ，后面括号中的 mean 表示这是一个平均值*/
- Requests per second: 13.45 [#/sec] (mean)
- /*大家最关心的指标之二，指的是用户平均请求等待时间
- 相当于 LR 中的 平均事务响应时间 ，后面括号中的 mean 表示这是一个平均值*/
- Time per request: 743.726 [ms] (mean)
- /*大家最关心的指标之三，指的是服务器平均请求处理时间
- Time per request: 74.373 [ms] (mean, across all concurrent requests)
- Transfer rate: 4.33 [Kbytes/sec] received
-
- Connection Times (ms)
- min mean[+/-sd] median max
- Connect: 129 163 245.3 145 3154
- Processing: 129 576 1510.8 147 11756
- Waiting: 129 567 1502.0 147 11756
- Total: 261 739 1543.7 294 11888
-
- Percentage of the requests served within a certain time (ms)
- 50% 294
- 66% 297
- 75% 304
- 80% 308
- 90% 1290
- 95% 3452
- 98% 7582
- 99% 7962
- 100% 11888 (longest request)
4、为了使结果更有对比性，我们将并发用户更改为100个进行压力测试，我这里只将三个指标贴出来。
- Requests per second: 190.95 [#/sec] (mean)
- Time per request: 523.694 [ms] (mean)
- Time per request: 5.237 [ms] (mean, across all concurrent requests)
5、将并发用户改为200个进行测试
- Requests per second: 186.00 [#/sec] (mean)
- Time per request: 1149.433 [ms] (mean)
- Time per request: 5.747 [ms] (mean, across all concurrent requests)
6、500个并发用户时的情况
- Requests per second: 180.99 [#/sec] (mean)
- Time per request: 2631.662 [ms] (mean)
- Time per request: 5.263 [ms] (mean, across all concurrent requests)
我们来分析下测试的结果，先对比下吞吐率，当并发用户的时候吞吐率最高为190 reqs/s,当并发用户数为200,500 吞吐率下降了，随之用户的等待时间更是明显增加了，已经有2s的等待时间了。这说明性能明显下降了。当然分析这个测试结果并不是说明51CTO的网站的并发用户只能在500左右，因为我是在服务器负荷的情况下就行测试的，这显然不能说明问题。另外我们在生产环境下测试的时候，最好能将测试结果做成报表，这样可以非常清晰地对比出问题来，好了，我该准备下，给上面提交一份我们公司网站的测试报告了。