Sysbench - 一种系统性能benchmark

SysBench是一款开源的、跨平台的、模块化的、多线程的性能测试工具，通过高负载地运行在数据库上，可以执行CPU/内存/线程/IO/数据库等方面的性能测试。用于评估操作系统的性能参数。

1          sysbench简介

Sysbench使得我们无需采用真正的复杂的数据库benchmark而获取系统的性能概况。

目前主要支持的数据库有mysql、oracle、postgresql。

可评估的系统性能参数有：

1.      Cpu性能

2.      内存分配与传输速度测试

3.      互斥体测试

4.      文件I/O性能

5.      线程调度性能

6.      数据库服务性能（OLTP基准）

本来sysbench主要是为mysql编写的benchmark。但是现在SysBench将进一步扩展，以支持更多的数据库backends，distributed benchmarks和第三方插件模块。

sysbench按照指定的数量开启线程，每个线程与mysql建立一个连接，每个线程不停地进行事务操作，打开事务，然后进行一些查询、更新、插入操作，提交事务，再开始新的事务；所有的sql只访问一个表-sbtest，是由sysbench的prepare命令建好的。其中的记录数，也是在prepare时指定好并创建的。测试结束，会有一个统计结果，包括例如每秒事务数、平均响应时间等等；

2          sysbench下载与安装

1.下载

下载地址：http://sourceforge.net/projects/sysbench/

获得sysbench-0.4.8.tar.gz源码。sysbench-0.4.8.tar.gz编译未通过。

2.安装

默认支持MySQL，如果需要测试Oracle/PostgreSQL，则在configure时需要加上–with-oracle或者–with-pgsql参数

[root@RedHat sysbench-0.4.8]# ./configure --without-mysql

[root@RedHat sysbench-0.4.8]# make

[root@RedHat sysbench-0.4.8]# make install

如果需要测试mysql，则需要配置参数选项：

./configure

--with-mysql-includes=/opt/mysql/include/mysql 表示mysql头文件目录

--with-mysql-libs=/opt/mysql/lib/mysql表示mysql模块目录

如股票缺少libtool请至ftp://ftp.gnu.org/gnu/libtool/下载

3          性能测试与结果分析

请查看sysbench帮助

[root@RedHat sysbench-0.4.8]# sysbench

Missing required command argument.

Usage:

sysbench [general-options]... --test=<test-name> [test-options]... command

General options:

--num-threads=N            number of threads to use [1]

--max-requests=N           limit for total number of requests [10000]

--max-time=N               limit for total execution time in seconds [0]

--thread-stack-size=SIZE   size of stack per thread [32K]

--init-rng=[on|off]        initialize random number generator [off]

--test=STRING              test to run

--debug=[on|off]           print more debugging info [off]

--validate=[on|off]        perform validation checks where possible [off]

--help=[on|off]            print help and exit

Compiled-in tests:

fileio - File I/O test

cpu - CPU performance test

memory - Memory functions speed test

threads - Threads subsystem performance test

mutex - Mutex performance test

oltp - OLTP test

Commands: prepare run cleanup help

See 'sysbench --test=<name> help' for a list of options for each test.

[root@RedHat sysbench-0.4.8]#

其中通用参数

参数名	参数描述
--num-threads=N	测试时使用的线程数，默认为1
--max-requests=N	请求数最大现在，默认10000
-max-time=N	执行时间最大限制，默认为0
--thread-stack-size=SIZE	每个线程的stack大小，默认是32K
--init-rng=[on|off]	是否初始化随机数生成器
--test=STRING	运行的测试
--debug=[on|off]	是否打印更多的debug信息
--validate=[on|off]	是否执行有效性检查
--help=[on|off]	打印帮助并退出

其中--test=STRING中的STRING可选项有：

STRING	描述
fileio	文件IO测试
cpu	CPU性能测试
memory	内存分配与传输速度测试
threads	线程子系统性能测试
mutex	互斥体性能测试
oltp	数据库OLTP测试

1. 文件IO测试

[root@RedHat test]# sysbench --test=fileio help

sysbench v0.4.8:  multi-threaded system evaluation benchmark

fileio options:

--file-num=N                  number of files to create [128]

--file-block-size=N           block size to use in all IO operations [16384]

--file-total-size=SIZE        total size of files to create [2G]

--file-test-mode=STRING       test mode {seqwr, seqrewr, seqrd, rndrd, rndwr,        rndrw}

--file-io-mode=STRING   file operations mode {sync,async,fastmmap,slowmmap} [sync]

--file-extra-flags=STRING  additional flags to use on opening files {sync,dsync,direct} []

--file-fsync-freq=N      do fsync() after this number of requests (0 - don't use fsync()) [100]

--file-fsync-all=[on|off]     do fsync() after each write operation [off]

--file-fsync-end=[on|off]     do fsync() at the end of test [on]

--file-fsync-mode=STRING which method to use for synchronization {fsync, fdatasync} [fsync]

--file-merged-requests=N merge at most this number of IO requests if poss ible (0- don't merge) [0]

--file-rw-ratio=N             reads/writes ratio for combined test [1.5]

[root@RedHat test]#

参数选项说明：

参数名	描述
--file-num=N	创建的文件数目，默认为128个
--file-block-size=N	所有IO操作中的block大小，默认是16384
--file-total-size=SIZE	创建的文件的总大小，默认为2G
--file-test-mode=STRING	测试的访问模式，可选一下6个参数：seqwr(顺序写)/seqrewr(顺序读写)/seqrd(顺序读)/ rndwr(随机写)/rndrw(随机读写)/rndrd(随机读)；
--file-io-mode=STRING	文件操作模式,有以下四种选择：sync,async,fastmmap,slowmmap默认为sync
--file-extra-flags=STRING	打开文件是的标识，有以下选择：sync,dsync,direct
--file-fsync-freq=N	在N个请求之后执行fsync()，默认值为100,0表示不执行fsync()
--file-fsync-all=[on|off]	是否每一次写操作之后均执行fsync()
--file-fsync-end=[on|off]	在测试结束时执行fsync()，默认开启
--file-fsync-mode=STRING	同步使用什么方法：fsync, fdatasync，默认为fsync
--file-merged-requests=N	尽可能的合并N个请求，0表示不合并，默认为0
--file-rw-ratio=N	读写比例默认为1.5：1

测试案例：

创建2个进程，创建的文件总大小为3G，文件读写模式为随机读。

可以获取读写速度~

[root@RedHat test]# sysbench --test=fileio --num-threads=2 --file-total-size=512M --file-test-mode=rndrw prepare （准备测试文件）

sysbench v0.4.8:  multi-threaded system evaluation benchmark

128 files, 4096Kb each, 512Mb total

Creating files for the test...

[root@RedHat test]# sysbench --test=fileio --num-threads=2 --file-total-size=512M --file-test-mode=rndrw run （运行测试）

sysbench v0.4.8:  multi-threaded system evaluation benchmark

Running the test with following options:

Number of threads: 2

Extra file open flags: 0

128 files, 4Mb each

512Mb total file size

Block size 16Kb

Number of random requests for random IO: 10000

Read/Write ratio for combined random IO test: 1.50

Periodic FSYNC enabled, calling fsync() each 100 requests.

Calling fsync() at the end of test, Enabled.

Using synchronous I/O mode

Doing random r/w test

Threads started!

Done.

Operations performed:  6000 Read, 4000 Write, 12800 Other = 22800 Total

Read 93.75Mb  Written 62.5Mb  Total transferred 156.25Mb  (3.2979Mb/sec)

211.07 Requests/sec executed

Test execution summary:

total time:                          47.3783s

total number of events:              10000

total time taken by event execution: 36.6912

per-request statistics:

min:                            0.0000s

avg:                            0.0037s

max:                            0.2687s

approx.  95 percentile:         0.0230s

Threads fairness:

events (avg/stddev):           5000.0000/180.00

execution time (avg/stddev):   18.3456/0.19

[root@RedHat test]# sysbench --test=fileio --num-threads=2 --file-total-size=512M --file-test-mode=rndrw cleanup （清除测试文件）

sysbench v0.4.8:  multi-threaded system evaluation benchmark

Removing test files...

[root@RedHat test]#

2. CPU性能测试

Sysbench采用寻找最大素数的方式测试CPU性能：

[root@RedHat ~]# sysbench --test=cpu help

sysbench v0.4.8:  multi-threaded system evaluation benchmark

cpu options:

--cpu-max-prime=N      upper limit for primes generator [10000]

[root@RedHat ~]#

参数解析

参数名	描述
--cpu-max-prime=N	素数的最大限制，默认为10000；

测试案例

[root@RedHat ~]# sysbench --test=cpu --cpu-max-prime=20000 run

sysbench v0.4.8:  multi-threaded system evaluation benchmark

Running the test with following options:

Number of threads: 1

Doing CPU performance benchmark

Threads started!

Done.

Maximum prime number checked in CPU test: 20000

Test execution summary:

total time:                          124.1761s

total number of events:              10000

total time taken by event execution: 123.3658

per-request statistics:

min:                            0.0051s

avg:                            0.0123s

max:                            0.1908s

approx.  95 percentile:         0.0171s

Threads fairness:

events (avg/stddev):           10000.0000/0.00

execution time (avg/stddev):   123.3658/0.00

[root@RedHat ~]#

3.内存性能测试

[root@RedHat ~]# sysbench --test=memory help

sysbench v0.4.8:  multi-threaded system evaluation benchmark

memory options:

--memory-block-size=SIZE    size of memory block for test [1K]

--memory-total-size=SIZE    total size of data to transfer [100G]

--memory-scope=STRING       memory access scope {global,local} [global]

--memory-hugetlb=[on|off]   allocate memory from HugeTLB pool [off]

--memory-oper=STRING        type of memory operations {read, write, none} [write]

--memory-access-mode=STRING memory access mode {seq,rnd} [seq]

[root@RedHat ~]#

参数名	描述
--memory-block-size=SIZE	内存块大小，默认为1K
--memory-total-size=SIZE	传输的数据量，默认为100G
--memory-scope=STRING	内存访问是局部local还是全局global，默认为global
--memory-hugetlb=[on|off]	从hugetlb中分配内存
--memory-oper=STRING	内存操作类型read, write, none，默认为写
--memory-access-mode=STRING	内存访问模式seq,rnd，默认为seq；

[root@RedHat test]# sysbench --test=memory --memory-block-size=4k --memory-total-size=64M run

4.线程性能测试

[root@RedHat ~]# sysbench --test=threads help

sysbench v0.4.8:  multi-threaded system evaluation benchmark

threads options:

--thread-yields=N      number of yields to do per request [1000]

--thread-locks=N       number of locks per thread [8]

[root@RedHat ~]#

参数名	描述
--thread-yields=N	内存块大小，默认为1K
--thread-locks=N	传输的数据量，默认为100G

测试案例

[root@RedHat test]# sysbench --test=threads --num-threads=64 --thread-yields=100 --thread-locks=2 run

sysbench v0.4.8:  multi-threaded system evaluation benchmark

Running the test with following options:

Number of threads: 64

Doing thread subsystem performance test

Thread yields per test: 100 Locks used: 2

Threads started!

Done.

Test execution summary:

total time:                          24.8159s

total number of events:              10000

total time taken by event execution: 1575.5529

per-request statistics:

min:                            0.0002s

avg:                            0.1576s

max:                            2.0173s

approx.  95 percentile:         0.5627s

Threads fairness:

events (avg/stddev):           156.2500/12.62

execution time (avg/stddev):   24.6180/0.08

[root@RedHat test]#

5互斥体性能测试

[root@RedHat ~]# sysbench --test=mutex help

sysbench v0.4.8:  multi-threaded system evaluation benchmark

mutex options:

--mutex-num=N        total size of mutex array [4096]

--mutex-locks=N      number of mutex locks to do per thread [50000]

--mutex-loops=N      number of empty loops to do inside mutex lock [10000]

[root@RedHat ~]#

参数名	描述
--mutex-num=N	互斥体数据大小，默认为4096
--mutex-locks=N	每个线程上互斥锁的数目，默认为50000；
--mutex-loops=N	互斥锁内空循环的数目，默认为10000

测试案例

[root@RedHat test]# sysbench --test=mutex --num-threads=16 --mutex-num=1024 --mutex-locks=10000 --mutex-loops=5000 run

sysbench v0.4.8:  multi-threaded system evaluation benchmark

Running the test with following options:

Number of threads: 16

Doing mutex performance test

Threads started!

Done.

Test execution summary:

total time:                          0.0482s

total number of events:              16

total time taken by event execution: 0.2271

per-request statistics:

min:                            0.0015s

avg:                            0.0142s

max:                            0.0312s

approx.  95 percentile:         0.0126s

Threads fairness:

events (avg/stddev):           1.0000/0.00

execution time (avg/stddev):   0.0142/0.01

[root@RedHat test]#

6 OLTP测试

不做分析了

参考资料：

http://sysbench.sourceforge.net/

http://blog.csdn.net/qiudakun/article/details/6891136

http://www.ningoo.net/html/2009/performance_test_tool_sysbench.html