Redis性能测试Redis-benchmark

=

最近排查redis的问题和优化，真费劲儿。

功夫都在细节处。

=

测试暂时用的脚本如下，还是能发现不少问题的。

-c并发数，

-n请求数，

-q仅仅输出简要结果避免太长，

-d以字节的形式指定 SET/GET 值的数据大小，

-r（Use random keys 且 指定key的长度，比如30）

./redis-benchmark -n 1000000 -q

./redis-benchmark -n 1000000 -q -r 30

./redis-benchmark -n 1000000 -q -r 50

./redis-benchmark -n 1000000 -q -r 80

./redis-benchmark -n 1000000 -q -r 30

./redis-benchmark -n 1000000 -q -r 30 -c 50

./redis-benchmark -n 1000000 -q -r 30 -c 50 -d 1024

./redis-benchmark -n 1000000 -q -r 30 -c 50 -d 2048

./redis-benchmark -n 1000000 -q -r 30 -d 398092

=

#首先看看标准文档，其实总比各种文要全。排查问题，很多还得弄的仔细些。

./redis-benchmark --help

Usage: redis-benchmark [-h <host>] [-p <port>] [-c <clients>] [-n <requests]> [-k <boolean>]

 

 -h <hostname>      Server hostname (default 127.0.0.1)

 -p <port>          Server port (default 6379)

 -s <socket>        Server socket (overrides host and port)

 -a <password>      Password for Redis Auth

 -c <clients>       Number of parallel connections (default 50)

 -n <requests>      Total number of requests (default 100000)

 -d <size>          Data size of SET/GET value in bytes (default 2)

 -dbnum <db>        SELECT the specified db number (default 0)

 -k <boolean>       1=keep alive 0=reconnect (default 1)

 -r <keyspacelen>   Use random keys for SET/GET/INCR, random values for SADD

  Using this option the benchmark will expand the string __rand_int__

  inside an argument with a 12 digits number in the specified range

  from 0 to keyspacelen-1. The substitution changes every time a command

  is executed. Default tests use this to hit random keys in the

  specified range.

 -P <numreq>        Pipeline <numreq> requests. Default 1 (no pipeline).

 -e                 If server replies with errors, show them on stdout.

                    (no more than 1 error per second is displayed)

 -q                 Quiet. Just show query/sec values

 --csv              Output in CSV format

 -l                 Loop. Run the tests forever

 -t <tests>         Only run the comma separated list of tests. The test

                    names are the same as the ones produced as output.

 -I                 Idle mode. Just open N idle connections and wait.

 

Examples:

 

 Run the benchmark with the default configuration against 127.0.0.1:6379:

   $ redis-benchmark

 

 Use 20 parallel clients, for a total of 100k requests, against 192.168.1.1:

   $ redis-benchmark -h 192.168.1.1 -p 6379 -n 100000 -c 20

 

 Fill 127.0.0.1:6379 with about 1 million keys only using the SET test:

   $ redis-benchmark -t set -n 1000000 -r 100000000

 

 Benchmark 127.0.0.1:6379 for a few commands producing CSV output:

   $ redis-benchmark -t ping,set,get -n 100000 --csv

 

 Benchmark a specific command line:

   $ redis-benchmark -r 10000 -n 10000 eval 'return redis.call("ping")' 0

 

 Fill a list with 10000 random elements:

   $ redis-benchmark -r 10000 -n 10000 lpush mylist __rand_int__

 

 On user specified command lines __rand_int__ is replaced with a random integer

 with a range of values selected by the -r option.

 

#例子

测试命令事例：

1、redis-benchmark -h 192.168.1.201 -p 6379 -c 100 -n 100000 
100个并发连接，100000个请求，检测host为localhost 端口为6379的redis服务器性能 

2、redis-benchmark -h 192.168.1.201 -p 6379 -q -d 100  

测试存取大小为100字节的数据包的性能

3、redis-benchmark -t set,lpush -n 100000 -q

只测试某些操作的性能

4、redis-benchmark -n 100000 -q script load "redis.call('set','foo','bar')"

只测试某些数值存取的性能

 

中文参考：redis-benchmark http://www.runoob.com/redis/redis-benchmarks.html

redis-benchmark使用方法总结  https://blog.csdn.net/xubo245/article/details/48320689 中文翻译：

 -h <hostname>      Server hostname (default 127.0.0.1)                    //服务主机名  

 -p <port>          Server port (default 6379)                             //服务器端口号  

 -s <socket>        Server socket (overrides host and port)                //socket，覆盖host和port  

 -a <password>      Password for Redis Auth                                //密码  

 -c <clients>       Number of parallel connections (default 50)            //并行请求的客户端数量，默认50  

 -n <requests>      Total number of requests (default 100000)              //请求的总量，默认100000  

 -d <size>          Data size of SET/GET value in bytes (default 2)        //set或get的数据尺寸，单位是字节，默认是2  

 -dbnum <db>        SELECT the specified db number (default 0)             //选择指定的数据号  

 -k <boolean>       1=keep alive 0=reconnect (default 1)                   //布尔量，1代表保持活跃，0代表重连  

 -r <keyspacelen>   Use random keys for SET/GET/INCR, random values for SADD //set/get/incr使用随机的keys,SADD使用随机值...  

  Using this option the benchmark will expand the string __rand_int__  

  inside an argument with a 12 digits number in the specified range  

  from 0 to keyspacelen-1. The substitution changes every time a command  

  is executed. Default tests use this to hit random keys in the  

  specified range.  

 -P <numreq>        Pipeline <numreq> requests. Default 1 (no pipeline).    //管道请求，默认是1，没管道（并行？）  

 -q                 Quiet. Just show query/sec values                       //安静，仅仅展示每秒的查询值  

 --csv              Output in CSV format                                    //输出按CSV格式  

 -l                 Loop. Run the tests forever  

 -t <tests>         Only run the comma separated list of tests. The test    //仅仅运行tests  

                    names are the same as the ones produced as output.  

 -I                 Idle mode. Just open N idle connections and wait.       //闲置模型，进打开N个闲置链接且等待  

=

=