Redis 性能测试
2020-09-21 本文已影响0人
_大叔_
服务器环境
| 参数 | linux指令 | 值 |
|---|---|---|
| 系统 | cat /etc/redhat-release | CentOS Linux release 7.8.2003 (Core) |
| 内存 | free -h | total：3.7G available：3.0G |
| CPU数量 | cat /proc/cpuinfo | cpu cores ： 2 |
| HZ | cat /proc/cpuinfo |grep MHz|uniq | cpu MHz : 2100.000 |
Redis 之 redis-benchmark
redis 提供了 redis-benchmark 工具便于我们做性能测试，可以使用 redis-benchmark- h 查看参数。
| 命令 | 描述 | 默认值 |
|---|---|---|
| -h <hostname> | 指定服务器主机名 | 127.0.0.1 |
| -p <port> | 指定服务器端口 | 6379 |
| -s <socket> | 指定服务器 socket | |
| -a <password> | Redis 认证密码 | |
| -c <clients> | 指定并发连接数 | 50 |
| -n <requests> | 指定请求数 | 10000 |
| -d <size> | 以字节（Byte）的形式指定 SET/GET 值的数据大小 | 3 |
| --dbnum <db> | 选择 Redis 数据库编号 | 0 |
| --threads <num> | 启用多线程模式 | redis6以上才有 |
| -k <boolean> | 是否保持连接 1=keep alive 0=reconnect | 1 |
| -r <keyspacelen> | SET/GET/INCR 使用随机 KEY ，SADD 使用随机值 | |
| -P <numreq> | 默认情况下，Redis 客户端一次请求只发起一个命令。通过 -P 参数，可以设置使用 pipelining功能，一次发起指定个请求，从而提升 QPS | 1 |
| -e | 如果 Redis Server 返回错误，是否将错误打印出来。默认情况下不打印，通过该参数开启 | |
| -q | 强制退出 redis。仅显示 query/sec 值 | |
| --csv | 以 CSV 格式输出 | |
| -l (L) | 生成循环，永久执行测试 | |
| -t <tests> | 指定需要测试的 Redis 命令，多个命令通过逗号分隔。想测试 SET/GET 命令，则可以 -t SET,GET 来指定 | |
| -I (i) | Idle 模式。仅打开 N 个 idle 连接并等待。 |
Docker Redis测试
单并发测试
测试命令
./redis-benchmark -c 1 -n 20000 -d 30720 -t set,get
测试结果每秒是 8000+的QPS
多并发测试
====== SET ======
20000 requests completed in 1.06 seconds
10 parallel clients
30720 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
18903.59 requests per second
====== GET ======
20000 requests completed in 1.30 seconds
10 parallel clients
30720 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
15360.98 requests per second
接下来测试 -P 命令
====== SET ======
20007 requests completed in 0.47 seconds
50 parallel clients
30720 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
42933.48 requests per second
====== GET ======
20002 requests completed in 0.44 seconds
50 parallel clients
30720 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
45253.39 requests per second
redis6 安装版测试
单并发测试
====== SET ======
200000 requests completed in 20.20 seconds
1 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
9900.01 requests per second
====== GET ======
200000 requests completed in 20.01 seconds
1 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
9995.50 requests per second
多并发测试
====== SET ======
200000 requests completed in 3.00 seconds
10 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
66622.25 requests per second
====== GET ======
200000 requests completed in 3.02 seconds
10 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
66137.57 requests per second
-P 测试
单并发测试
====== SET ======
200000 requests completed in 2.20 seconds
1 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
90991.81 requests per second
====== GET ======
200000 requests completed in 2.15 seconds
1 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
92980.01 requests per second
多并发测试
====== SET ======
200010 requests completed in 0.42 seconds
10 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
480793.25 requests per second
====== GET ======
200000 requests completed in 0.34 seconds
10 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
586510.25 requests per second
换服务器测试
| 参数 | linux指令 | 值 |
|---|---|---|
| 系统 | cat /etc/redhat-release | CentOS Linux release 7.7.1908 (Core) |
| 内存 | free -h | total：3.6G ，available：782M |
| CPU数量 | cat /proc/cpuinfo | cpu cores : 1 |
| HZ | cat /proc/cpuinfo |grep MHz|uniq | cpu MHz : 2500.002 |
Docker Redis
单并发测试
====== SET ======
20000 requests completed in 1.22 seconds
1 parallel clients
30720 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
16420.36 requests per second
====== GET ======
20000 requests completed in 1.42 seconds
1 parallel clients
30720 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
14094.43 requests per second
多并发测试
====== SET ======
20000 requests completed in 0.66 seconds
10 parallel clients
30720 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
30395.14 requests per second
====== GET ======
20000 requests completed in 0.81 seconds
10 parallel clients
30720 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
24721.88 requests per second
-P 测试
单并发测试
====== SET ======
20000 requests completed in 0.59 seconds
1 parallel clients
30720 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
33726.81 requests per second
====== GET ======
20000 requests completed in 0.50 seconds
1 parallel clients
30720 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
39682.54 requests per second
多并发测试
====== SET ======
20008 requests completed in 0.31 seconds
10 parallel clients
30720 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
64750.81 requests per second
====== GET ======
20017 requests completed in 0.27 seconds
10 parallel clients
30720 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
73591.91 requests per second
Redis 安装版
单并发
====== SET ======
200000 requests completed in 7.21 seconds
1 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
27735.40 requests per second
====== GET ======
200000 requests completed in 7.06 seconds
1 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
28312.57 requests per second
多并发
====== SET ======
200000 requests completed in 2.31 seconds
10 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
86730.27 requests per second
====== GET ======
200000 requests completed in 2.29 seconds
10 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
87221.98 requests per second
-P 测试
单并发
====== SET ======
200000 requests completed in 0.89 seconds
1 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
223964.16 requests per second
====== GET ======
200000 requests completed in 0.81 seconds
1 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
247524.75 requests per second
多并发
====== SET ======
200000 requests completed in 0.29 seconds
10 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
687285.19 requests per second
====== GET ======
200000 requests completed in 0.27 seconds
10 parallel clients
3 bytes payload
keep alive: 1
host configuration "save": 900 1 300 10 60 10000
host configuration "appendonly": no
multi-thread: no
751879.69 requests per second
明显可以看出 换了一台的服务器，CPU没有上一台多，可用内存剩余少，单单只是HZ不一样，但结果相差甚大。
其他运行测试，当一台32G的内存服务器，Redis跑到 7G 的时候 QPS 从 20W+ 一路下跌，跑到31G的时候QPS大约为 3W+。(这个大家可以自测 把 -d 值给大就行)
