启动命令

物理机

sysctl vm.overcommit_memory=1

echo never > /sys/kernel/mm/transparent_hugepage/enabled

/data/redis-6379/redis-4.0.14-centos7/redis-server redis.conf  \

  --logfile /data/redis-6379/redis-4.0.14-centos7/redis.log \

  --pidfile /var/run/redis_6379.pid \

  --daemonize yes \

  --port 6379 \

  --dir /data/redis-6379/redis-4.0.14-centos7/

k8s

kuberntes中的redis集群性能调优

现在做不到像那篇文章那样的调优，用的是默认配置。

结果

条件	set tps	get tps
windows本地	37878.79	39062.50
k8s集群-本地(kube-master2)	69930.07	71428.57
k8s集群-远程-跨主机 (kube-master2 -> hadoop2)	45662.10	50761.42
k8s集群-远程-跨主机 (hadoop2 -> kube-master2)	54347.82	56497.18
linux 物理机 本地 未优化内核参数 (hadoop3)	66225.17	66225.17
linux 物理机 本地 优化内核参数后 (hadoop3)	53475.93	63291.14
linux 物理机 远程 未优化内核参数 (hadoop2 -> hadoop3)	59171.60	59880.24
linux 物理机 远程 优化内核参数后 (hadoop2 -> hadoop3)	60606.06	62111.80

测试详情

参考：redis压力测试工具-----redis-benchmark

windows本地

C:\green\Redis-x64-4.0.2.2

λ redis-benchmark -h 127.0.0.1 -p 6379 -c 50 -n 10000 -t set

====== SET ======

  10000 requests completed in 0.26 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

71.24% <= 1 milliseconds

97.90% <= 2 milliseconds

99.97% <= 3 milliseconds

100.00% <= 3 milliseconds

37878.79 requests per second

C:\green\Redis-x64-4.0.2.2

λ redis-benchmark -h 127.0.0.1 -p 6379 -c 50 -n 10000 -t get

====== GET ======

  10000 requests completed in 0.26 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

84.28% <= 1 milliseconds

99.39% <= 2 milliseconds

99.98% <= 3 milliseconds

100.00% <= 3 milliseconds

39062.50 requests per second

k8s集群-本地(kube-master2)

/data # redis-benchmark -h 127.0.0.1 -p 6379 -c 50 -n 10000 -t set

====== SET ======

  10000 requests completed in 0.14 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

100.00% <= 0 milliseconds

69930.07 requests per second

/data # redis-benchmark -h 127.0.0.1 -p 6379 -c 50 -n 10000 -t get

====== GET ======

  10000 requests completed in 0.14 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

99.95% <= 1 milliseconds

100.00% <= 1 milliseconds

71428.57 requests per second

k8s集群-远程-跨主机 (kube-master2 -> hadoop2)

/data # redis-benchmark -h redis2 -p 6379 -c 50 -n 10000 -t set

====== SET ======

  10000 requests completed in 0.22 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

96.73% <= 1 milliseconds

97.92% <= 2 milliseconds

98.40% <= 3 milliseconds

98.79% <= 4 milliseconds

99.71% <= 5 milliseconds

100.00% <= 5 milliseconds

45662.10 requests per second

/data # redis-benchmark -h redis2 -p 6379 -c 50 -n 10000 -t get

====== GET ======

  10000 requests completed in 0.20 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

99.26% <= 1 milliseconds

99.51% <= 3 milliseconds

99.64% <= 4 milliseconds

100.00% <= 4 milliseconds

50761.42 requests per second

k8s集群-远程-跨主机 (hadoop2 -> kube-master2)

shijianjs@SHINELONPC:/mnt/c/java/git/k8s-test/practice/redis$ kubectl exec -it redis2-96bf85bdc-gjnd8 sh

/data # redis-benchmark -h redis1 -p 6379 -c 50 -n 10000 -t set

====== SET ======

  10000 requests completed in 0.18 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

96.78% <= 1 milliseconds

99.73% <= 2 milliseconds

100.00% <= 2 milliseconds

54347.82 requests per second

/data # redis-benchmark -h redis1 -p 6379 -c 50 -n 10000 -t get

====== GET ======

  10000 requests completed in 0.18 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

99.03% <= 1 milliseconds

100.00% <= 1 milliseconds

56497.18 requests per second

linux 物理机 本地 未优化内核参数 (hadoop3)

[root@hadoop3 redis-4.0.14-centos7]# ./redis-benchmark -h 127.0.0.1 -p 6379 -c 50 -n 10000 -t set

====== SET ======

  10000 requests completed in 0.15 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

100.00% <= 0 milliseconds

66225.17 requests per second

[root@hadoop3 redis-4.0.14-centos7]# ./redis-benchmark -h 127.0.0.1 -p 6379 -c 50 -n 10000 -t get

====== GET ======

  10000 requests completed in 0.15 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

100.00% <= 0 milliseconds

66225.17 requests per second

linux 物理机 本地 优化内核参数后 (hadoop3)

[root@hadoop3 redis-4.0.14-centos7]# ./redis-benchmark -h 127.0.0.1 -p 6379 -c 50 -n 10000 -t set

====== SET ======

  10000 requests completed in 0.19 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

95.98% <= 1 milliseconds

99.84% <= 2 milliseconds

100.00% <= 2 milliseconds

53475.93 requests per second

[root@hadoop3 redis-4.0.14-centos7]# ./redis-benchmark -h 127.0.0.1 -p 6379 -c 50 -n 10000 -t get

====== GET ======

  10000 requests completed in 0.16 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

100.00% <= 0 milliseconds

63291.14 requests per second

linux 物理机 远程 未优化内核参数 (hadoop2 -> hadoop3)

/data # redis-benchmark -h 10.60.44.54 -p 6379 -c 50 -n 10000 -t set

====== SET ======

  10000 requests completed in 0.17 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

100.00% <= 0 milliseconds

59171.60 requests per second

/data # redis-benchmark -h 10.60.44.54 -p 6379 -c 50 -n 10000 -t get

====== GET ======

  10000 requests completed in 0.17 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

99.52% <= 1 milliseconds

100.00% <= 1 milliseconds

59880.24 requests per second

linux 物理机 远程 优化内核参数后 (hadoop2 -> hadoop3)

/data # redis-benchmark -h 10.60.44.54 -p 6379 -c 50 -n 10000 -t set

====== SET ======

  10000 requests completed in 0.17 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

99.92% <= 1 milliseconds

100.00% <= 1 milliseconds

60606.06 requests per second

/data # redis-benchmark -h 10.60.44.54 -p 6379 -c 50 -n 10000 -t get

====== GET ======

  10000 requests completed in 0.16 seconds

  50 parallel clients

  3 bytes payload

  keep alive: 1

100.00% <= 0 milliseconds

62111.80 requests per second