压力测试工具sysbench

sysbench是一个模块化、跨平台、多线程基准测试工具，主要用于测试不同系统参数下的数据库负载情况，本文主要介绍0.4版本的使用。sysbench主要用于以下性能测试：

• 文件I/O性能
• 调度
• 内存分配和传输
• POSIX线程
• 数据库

安装

1.安装插件

yum install libtool -y

2.安装

./configure  --prefix=/usr/local/sysbench-0.4.12 --with-mysql-includes=/usr/local/mysql/include --with-mysql-libs=/usr/local/mysql/lib
make
make install
cp  -r sysbench/tests  /usr/local/sysbench-0.4.12
ln -s /usr/local/sysbench-0.4.12/bin/sysbench /usr/local/sysbench-0.4.12/sysbench

修改环境变量，在环境变量中加入

export LD_LIBRARY_PATH=/usr/local/mysql/lib

注意：如果安装目录中没有configure那么需要执行以下操作：

chmod +x autogen.sh

./autogen.sh

如果想要让 sysbench 支持 oracle /pgsql 的话，就需要在编译的时候加上参数
--with-oracle

或者

--with-pgsql

一般语法

prepare:用于文件IO和数据库OLTP测试的数据准备阶段。

run:性能测试阶段

cleanup:移除测试过程中产生的数据

help:显示帮助信息，获取--test帮助可以使用--test=name --help

通用命令选项

这部分命令参数的通用的，不管--test测试什么内容都可以使用这些通用的命令。可以执行sysbench --help了解各参数的具体解释

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]
  --forced-shutdown=STRING    amount of time to wait after --max-time before forcing shutdown [off]
  --thread-stack-size=SIZE    size of stack per thread [32K]
  --init-rng=[on|off]         initialize random number generator [off]
  --seed-rng=N                seed for random number generator, ignored when 0 [0]
  --tx-rate=N                 target transaction rate (tps) [0]
  --tx-jitter=N               target transaction variation, in microseconds [0]
  --report-interval=N         periodically report intermediate statistics with a specified interval in seconds. 0 disables intermediate reports [0]
  --report-checkpoints=[LIST,...]dump full statistics and reset all counters at specified points in time. The argument is a list of comma-separated values representing the amount of time in seconds elapsed from start of test when report checkpoint(s) must be performed. Report checkpoints are off by default. []
  --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
  --version=[on|off]          print version and exit

Log options:
  --verbosity=N      verbosity level {5 - debug, 0 - only critical messages} [4]

  --percentile=N      percentile rank of query response times to count [95]

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 version

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

主要的参数有：

--num-threads、

--max-requests、

--test

以下几个参数也经常会使用：

--max-time 最大的测试时长

--debug(开启debug可以显示更详细的每个线程的执行情况)

CPU测试

对CPU的性能测试通常有：1. 通过算质数；2计算圆周率等；sysbench使用的就是通过质数相加的测试。对CPU测试直接运行run即可

./sysbench --num-threads=12 --max-requests=10000 --debug=on --test=cpu --cpu-max-prime=20000 run

上面的测试是：12个线程执行1万条请求，每个请求执行质数相加到20000

thread测试

测试线程调度的性能，用于高负载下的线程性能测试。

--thread-yields=N      每个请求执行“lock/yield/unlock”循环的次数，默认1000

--thread-locks=N       每个线程的互斥锁，默认8个

./sysbench --num-threads=12 --max-requests=10000  --test=threads --thread-yields=100 --thread-locks=2 run

memory测试

内存分配测试，主要是针对不同的块大小进行内存的连续读写或者随机读写测试。

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]

1.8k顺序分配

./sysbench --num-threads=12 --max-requests=10000 --test=memory --memory-block-size=8K --memory-total-size=100G --memory-access-mode=seq run

[root@localhost sysbench-0.4.12]# ./sysbench --num-threads=12 --max-requests=10000 --test=memory --memory-block-size=8K --memory-total-size=100G --memory-access-mode=seq run
sysbench 0.4.12.10:  multi-threaded system evaluation benchmark

Running the test with following options:
Number of threads: 12
Random number generator seed is 0 and will be ignored

Doing memory operations speed test
Memory block size: 8K

Memory transfer size: 102400M

Memory operations type: write
Memory scope type: global
Threads started!
Done.

Operations performed: 13107200 (993893.95 ops/sec)

102400.00 MB transferred (7764.80 MB/sec)

General statistics:
    total time:                          13.1877s
    total number of events:              13107200
    total time taken by event execution: 91.9173
    response time:
         min:                                  0.00ms
         avg:                                  0.01ms
         max:                                  0.93ms
         approx.  95 percentile:               0.02ms

Threads fairness:
    events (avg/stddev):           1092266.6667/4629.30
    execution time (avg/stddev):   7.6598/0.02

报告：时间13S,7.7G/S

2.8k随机分配

./sysbench --num-threads=12 --max-requests=10000 --test=memory --memory-block-size=8K --memory-total-size=100G --memory-access-mode=rnd run

 

报告：12.6S,8G/s

3.16K顺序分配

./sysbench --num-threads=12 --max-requests=10000 --test=memory --memory-block-size=16K --memory-total-size=100G --memory-access-mode=seq run

 

报告：7S,13G/S

4.16K随机分配

./sysbench --num-threads=12 --max-requests=10000 --test=memory --memory-block-size=16K --memory-total-size=100G --memory-access-mode=rnd run

 

报告：6.4S,15G/s

总结：对于分配同样大小的内存，块月大分配的速率越快，随机分配比顺序分配的速度要快

文件io测试

./sysbench --test=fileio help

文件IO的测试主要用于测试IO的负载性能。主要的测试选项为--file-test-mode。还有几个可以关注的参数包括--file-block-size、--file-io-mode、--file-fsync-freq 、--file-rw-ratio

--file-num=N                   创建测试文件的数量,默认128个
  --file-block-size=N          block size大小,默认16K
  --file-total-size=SIZE       所有文件的总大小，默认2G
  --file-test-mode=STRING      测试类型 {seqwr(顺序写), seqrewr(顺序读写), seqrd(顺序读), rndrd(随机读), rndwr(随机写), rndrw(随机读写)}
  --file-io-mode=STRING         I/O模式,需要系统支持默认sync[sync(同步IO),async(异步IO),mmap()]
  --file-async-backlog=N        每个线程的异步操作队列数，默认128个，需要--file-io-mode=async;
  --file-extra-flags=STRING     additional flags to use on opening files {sync,dsync,direct} []
  --file-fsync-freq=N           当请求数达到多少时执行fsync()刷新,默认100,0代表过程中不执行fsync()
  --file-fsync-all=[on|off]     执行每一个写操作后执行fsync()刷新操作,默认关闭off
  --file-fsync-end=[on|off]     测试结束执行fsync()操作,默认开启on
  --file-fsync-mode=STRING      同步刷新方法,默认fsync {fsync, fdatasync}
  --file-merged-requests=N      合并指定数量的IO请求,0代表不合并,默认0
  --file-rw-ratio=N            读写比例，默认1.5/1

./sysbench --num-threads=12 --max-requests=10000  --test=fileio --file-total-size=3G --file-test-mode=rndrw prepare
./sysbench --num-threads=12 --max-requests=10000  --test=fileio --file-total-size=3G --file-test-mode=rndrw run
./sysbench --num-threads=12 --max-requests=10000  --test=fileio --file-total-size=3G --file-test-mode=rndrw clean

注意：如果要对比两台服务器的io性能，需要跑相同的线程。

互斥锁测试

互斥锁测试模拟所有线程在同一时刻并发运行

./sysbench  --num-threads=12  --test=mutex --mutex-num=1024 --mutex-locks=10000 --mutex-loops=10000 run

[root@localhost sysbench-0.4.12]# ./sysbench  --num-threads=12  --test=mutex --mutex-num=1024 --mutex-locks=10000 --mutex-loops=10000 run
sysbench 0.4.12.10:  multi-threaded system evaluation benchmark

Running the test with following options:
Number of threads: 12
Random number generator seed is 0 and will be ignored

Doing mutex performance test
Threads started!
Done.

General statistics:
    total time:                          0.0869s
    total number of events:              12
    total time taken by event execution: 1.0301
    response time:
         min:                                 83.86ms
         avg:                                 85.84ms
         max:                                 86.88ms
         approx.  95 percentile:              86.78ms

Threads fairness:
    events (avg/stddev):           1.0000/0.00
    execution time (avg/stddev):   0.0858/0.00

oltp测试

oltp是针对数据库的基准测试，例如每次对数据库进行优化后执行基准测试来测试不同的配置的tps。可以通过以下命令了解它的有关参数：

./sysbench --test=oltp help

--oltp-test-mode=STRING                    测试类型：simple(简单select测试),complex(事务测试),nontrx(非事务测试),sp(存储过程) ；默认complex
  --oltp-reconnect-mode=STRING             连接类型：session(每个线程到测试结束不重新连接),transaction(执行每个事务重新连接),query(每一个查询重新连接),random(随机)；默认 [session]
  --oltp-sp-name=STRING                    指定执行测试的存储过程名
  --oltp-read-only=[on|off]                仅执行select测试，默认关闭
  --oltp-avoid-deadlocks=[on|off]          更新过程中忽略死锁，默认[off]
  --oltp-skip-trx=[on|off]                 语句以bigin/commit开始结尾，默认[off]
  --oltp-range-size=N                      范围查询的范围大小，默认 [100]，例如begin 100 and 200
  --oltp-point-selects=N                   单个事务中select查询的数量，默认 [10]
  --oltp-use-in-statement=N                每个查询中主键查找(in 10个值)的数量，默认 [0]
  --oltp-simple-ranges=N                   单个事务中执行范围查询的数量(SELECT c  FROM sbtest WHERE id BETWEEN  N AND  M)，默认[1]
  --oltp-sum-ranges=N                      单个事务中执行范围sum查询的数量，默认 [1]
  --oltp-order-ranges=N                    单个事务中执行范围order by查询的数量，默认[1]
  --oltp-distinct-ranges=N                 单个事务中执行范围distinct查询的数量，默认[1]
  --oltp-index-updates=N                   单个事务中执行索引更新的操作的数量，默认[1]
  --oltp-non-index-updates=N               单个事务中执行非索引更新操作的数量，默认[1]
  --oltp-nontrx-mode=STRING                指定单独非事务测试类型进行测试，默认select {select, update_key, update_nokey, insert, delete} [select]
  --oltp-auto-inc=[on|off]                 id列默认自增，默认[on]
  --oltp-connect-delay=N                   指定每一次重新连接延时的时长，默认1秒 [10000]
  --oltp-user-delay-min=N                  minimum time in microseconds to sleep after each request [0]
  --oltp-user-delay-max=N                  maximum time in microseconds to sleep after each request [0]
  --oltp-table-name=STRING                 指定测试的表名，默认[sbtest]
  --oltp-table-size=N                      指定表的记录大小，默认[10000]
  --oltp-dist-type=STRING                  随机数分布状态。uniform(均匀分布)、gauss(高斯分布)、special(特殊分布)，默认 [special]
  --oltp-dist-iter=N                       number of iterations used for numbers generation [12]
  --oltp-dist-pct=N                        启用百分比特殊分布，默认 [1]
  --oltp-dist-res=N                        special 百分比[75]
  --oltp-point-select-mysql-handler=[on|off] Use MySQL HANDLER for point select [off]
  --oltp-point-select-all-cols=[on|off]    select查询测试时select所有列，默认[off]
  --oltp-secondary=[on|off]                索引不是主键索引而是二级索引，默认[off]
  --oltp-num-partitions=N                  指定表分区的数量，默认 [0]
  --oltp-num-tables=N                      指定测试表的数量，默认[1]
General database options:
  --db-driver=STRING  指定测试数据库类型，默认mysql
  --db-ps-mode=STRING prepared statements usage mode {auto, disable} [auto]

mysql options:
  --mysql-host=[LIST,...]       MySQL server host [localhost]
  --mysql-port=N                MySQL server port [3306]
  --mysql-socket=STRING         MySQL socket
  --mysql-user=STRING           MySQL user [sbtest]
  --mysql-password=STRING       MySQL password []
  --mysql-db=STRING             MySQL database name [sbtest]
  --mysql-table-engine=STRING   storage engine to use for the test table {myisam,innodb,bdb,heap,ndbcluster,federated} [innodb]
  --mysql-engine-trx=STRING     whether storage engine used is transactional or not {yes,no,auto} [auto]
  --mysql-ssl=[on|off]          use SSL connections, if available in the client library [off]
  --myisam-max-rows=N           max-rows parameter for MyISAM tables [1000000]
  --mysql-create-options=STRING additional options passed to CREATE TABLE []

oltp测试主要会有以下相关参数的测试,,其它相关参数默认即可，有需求也可以自定义：

--mysql-engine-trx=STRING     指定不同的存储引擎测试。
--oltp-test-mode=STRING       测试类型：simple(简单select测试),complex(事务测试),nontrx(非事务测试),sp(存储过程) ；默认complex
--oltp-sp-name=STRING         指定存储过程进行语句测试
--oltp-table-size=N           指定表的记录大小，默认[10000]
--oltp-num-tables=N           指定测试表的数量，默认[1]

需要先创建好测试数据库sbtest。

1.事务测试，测试12个线程执行1万条请求，10个表，每个表大小100W

./sysbench --num-threads=12 --max-requests=100000  --test=oltp --mysql-user=root --mysql-password=root --oltp-test-mode=complex --mysql-db=sbtest --oltp-table-size=1000000 --oltp-num-tables=10 prepare

./sysbench --num-threads=12 --max-requests=100000  --test=oltp --mysql-user=root --mysql-password=root --oltp-test-mode=complex --mysql-db=sbtest --oltp-table-size=1000000 --oltp-num-tables=10 run

./sysbench --num-threads=12 --max-requests=100000  --test=oltp --mysql-user=root --mysql-password=root --oltp-test-mode=complex --mysql-db=sbtest --oltp-table-size=1000000 --oltp-num-tables=10 cleanup

[root@localhost sysbench-0.4.12]# ./sysbench --num-threads=12 --max-requests=100000  --test=oltp --mysql-user=root --mysql-password=root --oltp-test-mode=complex --mysql-db=sbtest --oltp-table-size=1000000 --oltp-num-tables=10 run
sysbench 0.4.12.10:  multi-threaded system evaluation benchmark

No DB drivers specified, using mysql
Running the test with following options:
Number of threads: 12
Random number generator seed is 0 and will be ignored

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
Using 10 test tables
Threads started!
Done.

OLTP test statistics:
    queries performed:
        read:                            1400910
        write:                           500325
        other:                           200130
        total:                           2101365
    transactions:                        100065 (791.94 per sec.)
    deadlocks:                           0      (0.00 per sec.)
    read/write requests:                 1901235 (15046.89 per sec.)
    other operations:                    200130 (1583.88 per sec.)

General statistics:
    total time:                          126.3540s
    total number of events:              100065
    total time taken by event execution: 1513.9202
    response time:
         min:                                  5.82ms
         avg:                                 15.13ms
         max:                               2352.66ms
         approx.  95 percentile:              18.00ms

Threads fairness:
    events (avg/stddev):           8338.7500/207.

报告：执行当前测试花费126S，TPS:791/S，RQ:15046/S，95%的请求花费18毫秒

事务测试，每个事务包含如下语句：

  Point queries:
SELECT c FROM sbtest WHERE id= N
  Range queries:
SELECT c FROM sbtest WHERE id BETWEEN  N AND  M
  Range SUM() queries:
SELECT SUM(K) FROM sbtest WHERE id BETWEEN  N and  M
  Range ORDER BY queries:
SELECT c FROM sbtest WHERE id between  N and  M ORDER BY c
  Range DISTINCT queries:
SELECT DISTINCT c FROM sbtest WHERE id BETWEEN  N and  M ORDER BY
c
  UPDATEs on index column:
UPDATE sbtest SET k=k+1 WHERE id= N
  UPDATEs on non-index column:
UPDATE sbtest SET c= N WHERE id= M
  DELETE queries:
DELETE FROM sbtest WHERE id= N
  INSERT queries:
INSERT INTO sbtest VALUES ( ...

 非事务测试，执行语句如下

  Point queries:
SELECT pad FROM sbtest WHERE id= N
  UPDATEs on index column:
UPDATE sbtest SET k=k+1 WHERE id= N
  UPDATEs on non-index column:
UPDATE sbtest SET c= N WHERE id= M
  DELETE queries:
DELETE FROM sbtest WHERE id= N
  The generated row IDs are unique over each test run, so no row
is deleted twice.
  INSERT queries:
INSERT INTO sbtest (k, c, pad) VALUES( N ,  M ,  S )

总结

sysbench是使用最广泛的基准压测工具，功能也很齐全报告也非常的详细。