sysbench是一款开源的多线程性能测试工具,可以执行CPU/内存/线程/IO/数据库等方面的性能测试,数据库目前支持MySQL/Oracle/PostgreSQL,下面我们来看看它的安装与测试.
如果评测一台mysql数据库的压力,可以使用sysbench来测试,具体的操作出下,先安装sysbench工具,安装操作如下.
安装环境
CentOS release 5.4 (Final),MySQL 5.1.40 MySQL_HOME=/usr/local/mysql/,Sysbench 0.4.12
安装步骤
1.去http://sourceforge.net/projects/sysbench/下载最新版本的sysbench 0.4.12
2.解压缩sysbench-0.4.12.tar.gz
tar --zxvf sysbench-0.4.12.tar.gz
3.进入解压缩包 sysbench-0.4.12,并执行脚本autogen.sh
cd sysbench-0.4.12
./autogen.sh
4.关键的三部曲:configure&make&make install
首先是./configure命令，sysbench默认是支持MySQL的benchmarking的,如果不加任何选项则要求保证MySQL的安装路径都是默认的标准路径,headfile位于/usr/include目录下,libraries位于/usr/lib/目录下,因为我的MySQL是源码编译安装的,安装路径是放在/usr/local/mysql下,所以这里要添加相应的选项命令:
- ./configure --prefix=/usr/local/mysql --with-mysql=/usr/local/mysql --with-mysql-includes=/usr/local/mysql/include/mysql/ --with-mysql-libs=/usr/local/mysql/lib/mysql/
注意:经反复测试这里在编译时要将路径写到最后的include/mysql,lib/mysql
--with-mysql-includes=/usr/local/mysql/include/mysql/
--with-mysql-libs=/usr/local/mysql/lib/mysql/
因为网上好多资料都没有提到这一层,在编译时总是编译不过去,这里浪费了好多精力.
make
make install
安装就可以了,下面再来看测试.
sysbench数据库的测试案例:
第一步:准备数据库,准备在mysql数据库创建1000000数据量,存储引擎是innodb.
- bin/sysbench --test=oltp / --测试类型数据库oltp
- --mysql-host=127.0.0.1 / --mysql主机
- --mysql-port=3421 / --mysql端口
- --mysql-user=root / --mysql登陆用户
- --mysql-password=$password / --mysql密码
- --mysql-db=test / --mysql测试数据库名
- --mysql-socket=/var/mysql.sock / --socket位置
- --oltp-table-name=test / --mysql测试表名
- --mysql-table-engine=innodb / --mysql表使用存储引擎
- --oltp-table-size=1000000 prepare --表数据量
第二步:开始测试.
- bin/sysbench --test=oltp /
- --mysql-host=127.0.0.1 / --mysql主机
- --mysql-port=3421 / --mysql端口
- --mysql-user=root / --mysql登陆用户
- --mysql-password=$password / --mysql密码
- --mysql-db=test / --mysql测试数据库名
- --mysql-socket=/var/mysql.sock / --socket位置
- --oltp-table-name=test / --mysql测试表名
- --mysql-table-engine=innodb / --mysql表使用存储引擎
- --max-requests=100000 / --测试请求次数
- --max-time=600 / --测试最长时间
- --num-threads=100 run --并发线程数
- //测试结果：
- sysbench 0.4.12: multi-threaded system evaluation benchmark
- No DB drivers specified, using mysql
- Running the test with following options:
- Number of threads: 100
- Doing OLTP test.
- Running mixed OLTP test
- Using Special distribution (12 iterations, 1 pct of values are returned in 75 pct cases)
- Using "BEGIN" for starting transactions
- Using auto_inc on the id column
- Maximum number of requests for OLTP test is limited to 100000
- Threads started!
- Done.
- OLTP test statistics:
- queries performed:
- read: 1401582
- write: 500563
- other: 200225
- total: 2102370
- transactions: 100112 (3446.37 per sec.)
- deadlocks: 1 (0.03 per sec.)
- read/write requests: 1902145 (65481.66 per sec.)
- other operations: 200225 (6892.78 per sec.)
- Test execution summary:
- total time: 29.0485s
- total number of events: 100112
- total time taken by event execution: 2898.7554
- per-request statistics:
- min: 3.52ms
- avg: 28.96ms
- max: 187.11ms
- approx. 95 percentile: 46.05ms
- Threads fairness:
- events (avg/stddev): 1001.1200/10.80
- execution time (avg/stddev): 28.9876/0.03
第三步:清理测试数据.
- bin/sysbench --test=oltp /
- --mysql-host=127.0.0.1 / --mysql主机
- --mysql-port=3421 / --mysql端口
- --mysql-user=root / --mysql登陆用户
- --mysql-password=$password / --mysql密码
- --mysql-db=test / --mysql测试数据库名
- --mysql-socket=/var/mysql.sock / --socket位置
- --oltp-table-name=test / --mysql测试表名
- --mysql-table-engine=innodb / --mysql表使用存储引擎
- cleanup
在版本0.5中支持自定义语句,可以参考下面的链接.
http://dba.stackexchange.com/questions/39221/stress-test-mysql-with-queries-captured-with-general-log-in-mysql
3.测试案例,表属性compress对insert速度的影响.
第一步准备:创建表结构:
- bin/sysbench --test=oltp / --测试类型数据库oltp
- --mysql-host=127.0.0.1 / --mysql主机
- --mysql-port=3421 / --mysql端口
- --mysql-user=root / --mysql登陆用户
- --mysql-password=$password / --mysql密码
- --mysql-db=test / --mysql测试数据库名
- --mysql-socket=/var/mysql.sock / --socket位置
- --oltp-table-name=sbtest / --mysql测试表名
- --mysql-table-engine=innodb / --mysql表使用存储引擎
- --oltp-table-size=1 prepare --表数据量
将表名修改为uncompressed
rename table sbtest to uncompressed;
再次创建表sbtest,并将表名修改为compressed
rename table sbtest to compressed;
将表改为compressed
alter table compressed ROW_FORMAT=compressed;
第二步:向表uncompressed表中添加记录
- bin/sysbench --test=oltp /
- --mysql-host=127.0.0.1 / --mysql主机
- --mysql-port=3421 / --mysql端口
- --mysql-user=root / --mysql登陆用户
- --mysql-password=$password / --mysql密码
- --mysql-db=test / --mysql测试数据库名
- --mysql-socket=/var/mysql.sock / --socket位置
- --oltp-table-name=uncompressed / --mysql测试表名
- --mysql-table-engine=innodb / --mysql表使用存储引擎
- --oltp-nontrx-mode=insert /
- --oltp-test-mode=nontrx /
- --max-requests=1000000 /
- --max-time=600 /
- --num-threads=100 run
- --测试结果：
- OLTP test statistics:
- queries performed:
- read: 0
- write: 1001110
- other: 0
- total: 1001110
- transactions: 1001110 (13617.75 per sec.)
- deadlocks: 0 (0.00 per sec.)
- read/write requests: 1001110 (13617.75 per sec.)
- other operations: 0 (0.00 per sec.)
- Test execution summary:
- total time: 73.5151s
- total number of events: 1001110
- total time taken by event execution: 7346.0036
- per-request statistics:
- min: 0.10ms
- avg: 7.34ms
- max: 239.05ms
- approx. 95 percentile: 22.00ms
- Threads fairness:
- events (avg/stddev): 10011.1000/86.80
- execution time (avg/stddev): 73.4600/0.00
说明:总共消耗了73s,每秒达到13617个事务.
第三步:向表compressed表中添加记录:
- bin/sysbench --test=oltp /
- --mysql-host=127.0.0.1 / --mysql主机
- --mysql-port=3421 / --mysql端口
- --mysql-user=root / --mysql登陆用户
- --mysql-password=$password / --mysql密码
- --mysql-db=test / --mysql测试数据库名
- --mysql-socket=/var/mysql.sock / --socket位置
- --oltp-table-name=compressed / --mysql测试表名
- --mysql-table-engine=innodb / --mysql表使用存储引擎
- --oltp-nontrx-mode=insert /
- --oltp-test-mode=nontrx /
- --max-requests=1000000 /
- --max-time=600 /
- --num-threads=100 run
- --测试结果：
- OLTP test statistics:
- queries performed:
- read: 0
- write: 1000515
- other: 0
- total: 1000515
- transactions: 1000515 (2313.85 per sec.)
- deadlocks: 0 (0.00 per sec.)
- read/write requests: 1000515 (2313.85 per sec.)
- other operations: 0 (0.00 per sec.)
- Test execution summary:
- total time: 432.4036s
- total number of events: 1000515
- total time taken by event execution: 43229.6698
- per-request statistics:
- min: 0.10ms
- avg: 43.21ms
- max: 730.32ms
- approx. 95 percentile: 133.46ms
- Threads fairness:
- events (avg/stddev): 10005.1500/109.30
- execution time (avg/stddev): 432.2967/0.01
说明:总共消耗了432s,每秒处理2313事务数,和uncompressed的速度相差6倍.
4.测试案例,表属性compress对复合操作速度的影响.
第一步:准备数据,和第3项一样准备.
第二步:针对compressed表进行复合操作测试.
- bin/sysbench --test=oltp /
- --mysql-host=127.0.0.1 / --mysql主机
- --mysql-port=3421 / --mysql端口
- --mysql-user=root / --mysql登陆用户
- --mysql-password=$password / --mysql密码
- --mysql-db=test / --mysql测试数据库名
- --mysql-socket=/var/mysql.sock / --socket位置
- --oltp-table-name=compressed / --mysql测试表名
- --mysql-table-engine=innodb / --mysql表使用存储引擎
- --max-requests=1000000 /
- --max-time=600 /
- --num-threads=100 run
- OLTP test statistics:
- queries performed:
- read: 14012236
- write: 5004356
- other: 2001743
- total: 21018335
- transactions: 1000869 (3145.37 per sec.)
- deadlocks: 5 (0.02 per sec.)
- read/write requests: 19016592 (59762.20 per sec.)
- other operations: 2001743 (6290.75 per sec.)
- Test execution summary:
- total time: 318.2043s
- total number of events: 1000869
- total time taken by event execution: 31800.5571
- per-request statistics:
- min: 3.19ms
- avg: 31.77ms
- max: 344.54ms
- approx. 95 percentile: 53.37ms
- Threads fairness:
- events (avg/stddev): 10008.6900/32.50
- execution time (avg/stddev): 318.0056/0.01
说明:共消耗时间318s,每秒3145个事务.
第三步:针对uncompressed表进行复合操作测试.
- bin/sysbench --test=oltp /
- --mysql-host=127.0.0.1 / --mysql主机
- --mysql-port=3421 / --mysql端口
- --mysql-user=root / --mysql登陆用户
- --mysql-password=$password / --mysql密码
- --mysql-db=test / --mysql测试数据库名
- --mysql-socket=/var/mysql.sock / --socket位置
- --oltp-table-name=uncompressed / --mysql测试表名
- --mysql-table-engine=innodb / --mysql表使用存储引擎
- --oltp-nontrx-mode=insert /
- --oltp-test-mode=nontrx /
- --max-requests=1000000 /
- --max-time=600 /
- --num-threads=100 run
- OLTP test statistics:
- queries performed:
- read: 14013370
- write: 5004769
- other: 2001908
- total: 21020047
- transactions: 1000953 (3389.22 per sec.)
- deadlocks: 2 (0.01 per sec.)
- read/write requests: 19018139 (64395.20 per sec.)
- other operations: 2001908 (6778.44 per sec.)
- Test execution summary:
- total time: 295.3347s
- total number of events: 1000953
- total time taken by event execution: 29512.3204
- per-request statistics:
- min: 3.37ms
- avg: 29.48ms
- max: 157.93ms
- approx. 95 percentile: 48.19ms
- Threads fairness:
- events (avg/stddev): 10009.5300/30.59
- execution time (avg/stddev): 295.1232/0.02
说明:共消耗时间295s,每秒3389个事务,在复合操作中,compressed的表会快一些.
新闻热点
疑难解答
