828华为云征文｜华为云Flexus云服务器X实例下的Redis性能基准测试

微微CMS

已于 2024-09-25 22:19:22 修改

阅读量740

点赞数 6

文章标签：华为云服务器 redis

于 2024-09-25 22:13:09 首次发布

本文链接：https://blog.csdn.net/weixin_48810756/article/details/142534264

版权

Flexus云服务器X实例介绍

华为云Flexus云服务是专为中小企业和开发者设计的卓越云解决方案，提供出色的开箱即用体验和显著的性能提升。其中，Flexus云服务器X实例作为这一服务中的杰出代表，展现了卓越的能力。

Flexus云服务器X实例是针对中小企业和开发者的下一代灵活计算云服务器，能够智能地适应不同业务负载，特别适用于电商直播、企业网站、开发测试环境、游戏服务器及音视频服务等中低负载场景。与Flexus应用服务器L实例相比，X实例不仅提供了丰富的公共镜像选项，还支持灵活的vCPU和内存配置，并能智能调整资源，以满足更高的负载需求

华为云Flexus云服务器X实例性能强悍，内嵌智能应用调优算法，底层多重调优加速，凝结华为技术专家多年经验，让Flexus X实例的基础模式GeekBench单核及多核跑分可达业界同规格独享型实例1.6倍；Flexus X实例的性能模式在相同性能和可靠性SLA下可超过业界C系/G系/R系，以及S系旗舰型云主机的性能。

测评关注点

华为云的Flexus提供了开箱即用体验，独有X-Turbo加速、大模型底层智能调度等黑科技，Flexus X实例独家提供业务关键应用的加速功能。在Flexus X实例上部署的MySQL、Redis、Nginx，性能最高可达业界同规格独享型实例的6倍（MySQL性能），长时运行可达2倍。

本次测评就拿Flexus X实例提供的Redis来开刀进行测试，看看基准测试性能表现如何。

关于Redis Benchmark工具

Redis Benchmark（redis-benchmark）是Redis官方提供的一个用于测试 Redis 性能的工具。它可以帮助开发者和运维人员评估 Redis 服务器的响应速度和处理能力。

主要功能：

性能测试：通过模拟多种类型的 Redis 操作（如 GET、SET、LPUSH 等），评估 Redis 在不同负载下的性能表现。
并发测试：支持指定并发连接数，模拟多个客户端同时对 Redis 进行操作，测试其在高并发情况下的表现。
结果分析：提供请求的响应时间、每秒请求数等统计信息，便于用户了解 Redis 的性能瓶颈

Redis官方出品的基准性能测试工具，想必大家都不必担心基准测试的公平性吧！

测试环境

操作系统：Huawei Cloud EulerOS 2.0 标准版 64位

CPU：4核心

内存：12G

硬盘：100G
Redis版本 v6.2.7

说明：使用了Flexus X实例提供的Redis加速

开始测试

GET/SET 基准测试

直接来30W的总数据，1000个客户端写入测试看看效果

[root@flexusx-45d1 ~]# redis-benchmark  -c 1000 -n 300000 -t get,set

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

  300000 requests completed in 1.74 seconds

  1000 parallel clients

  3 bytes payload

  keep alive: 1

  host configuration "save": 3600 1 300 100 60 10000

  host configuration "appendonly": no

  multi-thread: no



Latency by percentile distribution:

0.000% <= 1.031 milliseconds (cumulative count 2)

50.000% <= 2.911 milliseconds (cumulative count 157216)

75.000% <= 2.975 milliseconds (cumulative count 229832)

87.500% <= 3.031 milliseconds (cumulative count 264004)

93.750% <= 3.103 milliseconds (cumulative count 281690)

96.875% <= 3.191 milliseconds (cumulative count 290866)

98.438% <= 3.271 milliseconds (cumulative count 295405)

99.219% <= 3.447 milliseconds (cumulative count 297675)

99.609% <= 4.239 milliseconds (cumulative count 298833)

99.805% <= 4.871 milliseconds (cumulative count 299420)

99.902% <= 5.215 milliseconds (cumulative count 299712)

99.951% <= 5.511 milliseconds (cumulative count 299857)

99.976% <= 5.647 milliseconds (cumulative count 299929)

99.988% <= 5.743 milliseconds (cumulative count 299964)

99.994% <= 5.855 milliseconds (cumulative count 299982)

99.997% <= 6.655 milliseconds (cumulative count 299991)

99.998% <= 6.679 milliseconds (cumulative count 299996)

99.999% <= 6.695 milliseconds (cumulative count 299998)

100.000% <= 6.783 milliseconds (cumulative count 299999)

100.000% <= 6.791 milliseconds (cumulative count 300000)

100.000% <= 6.791 milliseconds (cumulative count 300000)



Cumulative distribution of latencies:

0.000% <= 0.103 milliseconds (cumulative count 0)

0.010% <= 1.103 milliseconds (cumulative count 31)

0.023% <= 1.207 milliseconds (cumulative count 68)

0.032% <= 1.303 milliseconds (cumulative count 96)

0.043% <= 1.407 milliseconds (cumulative count 128)

0.054% <= 1.503 milliseconds (cumulative count 162)

0.066% <= 1.607 milliseconds (cumulative count 199)

0.077% <= 1.703 milliseconds (cumulative count 232)

0.083% <= 1.807 milliseconds (cumulative count 248)

0.096% <= 1.903 milliseconds (cumulative count 289)

0.119% <= 2.007 milliseconds (cumulative count 357)

0.140% <= 2.103 milliseconds (cumulative count 420)

93.897% <= 3.103 milliseconds (cumulative count 281690)

99.555% <= 4.103 milliseconds (cumulative count 298664)

99.878% <= 5.103 milliseconds (cumulative count 299633)

99.997% <= 6.103 milliseconds (cumulative count 299990)

100.000% <= 7.103 milliseconds (cumulative count 300000)



Summary:

  throughput summary: 172413.80 requests per second

  latency summary (msec):

          avg       min       p50       p95       p99       max

        2.921     1.024     2.911     3.135     3.367     6.791

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

  300000 requests completed in 1.75 seconds

  1000 parallel clients

  3 bytes payload

  keep alive: 1

  host configuration "save": 3600 1 300 100 60 10000

  host configuration "appendonly": no

  multi-thread: no



Latency by percentile distribution:

0.000% <= 1.007 milliseconds (cumulative count 1)

50.000% <= 2.935 milliseconds (cumulative count 156417)

75.000% <= 2.999 milliseconds (cumulative count 232359)

87.500% <= 3.039 milliseconds (cumulative count 265842)

93.750% <= 3.079 milliseconds (cumulative count 283317)

96.875% <= 3.127 milliseconds (cumulative count 290731)

98.438% <= 3.215 milliseconds (cumulative count 295570)

99.219% <= 3.431 milliseconds (cumulative count 297679)

99.609% <= 4.639 milliseconds (cumulative count 298830)

99.805% <= 7.271 milliseconds (cumulative count 299416)

99.902% <= 9.535 milliseconds (cumulative count 299708)

99.951% <= 10.831 milliseconds (cumulative count 299854)

99.976% <= 11.431 milliseconds (cumulative count 299927)

99.988% <= 11.799 milliseconds (cumulative count 299964)

99.994% <= 11.991 milliseconds (cumulative count 299982)

99.997% <= 12.087 milliseconds (cumulative count 299991)

99.998% <= 12.151 milliseconds (cumulative count 299996)

99.999% <= 12.183 milliseconds (cumulative count 299998)

100.000% <= 12.191 milliseconds (cumulative count 299999)

100.000% <= 12.207 milliseconds (cumulative count 300000)

100.000% <= 12.207 milliseconds (cumulative count 300000)



Cumulative distribution of latencies:

0.000% <= 0.103 milliseconds (cumulative count 0)

0.000% <= 1.007 milliseconds (cumulative count 1)

0.013% <= 1.103 milliseconds (cumulative count 39)

0.024% <= 1.207 milliseconds (cumulative count 73)

0.038% <= 1.303 milliseconds (cumulative count 114)

0.058% <= 1.407 milliseconds (cumulative count 175)

0.081% <= 1.503 milliseconds (cumulative count 242)

0.089% <= 1.607 milliseconds (cumulative count 266)

0.095% <= 1.703 milliseconds (cumulative count 285)

0.098% <= 1.807 milliseconds (cumulative count 294)

0.102% <= 1.903 milliseconds (cumulative count 305)

0.106% <= 2.007 milliseconds (cumulative count 317)

0.113% <= 2.103 milliseconds (cumulative count 338)

95.835% <= 3.103 milliseconds (cumulative count 287505)

99.459% <= 4.103 milliseconds (cumulative count 298376)

99.695% <= 5.103 milliseconds (cumulative count 299085)

99.753% <= 6.103 milliseconds (cumulative count 299258)

99.798% <= 7.103 milliseconds (cumulative count 299393)

99.843% <= 8.103 milliseconds (cumulative count 299528)

99.882% <= 9.103 milliseconds (cumulative count 299646)

99.923% <= 10.103 milliseconds (cumulative count 299769)

99.962% <= 11.103 milliseconds (cumulative count 299885)

99.997% <= 12.103 milliseconds (cumulative count 299992)

100.000% <= 13.103 milliseconds (cumulative count 300000)



Summary:

  throughput summary: 171135.19 requests per second

  latency summary (msec):

          avg       min       p50       p95       p99       max

        2.942     1.000     2.935     3.095     3.287    12.207

GET/SET 开启多线程的基准测试

在上面的基准测试中，还没有使用多线程的能力，接下来开启使用多线程来进行GET/SET的测试，看看QPS在多少。

[root@flexusx-45d1 ~]# redis-benchmark --threads 4 -c 1000 -n 300000 -t get,set

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

  300000 requests completed in 1.26 seconds

  1000 parallel clients

  3 bytes payload

  keep alive: 1

  host configuration "save": 3600 1 300 100 60 10000

  host configuration "appendonly": no

  multi-thread: yes

  threads: 4



Latency by percentile distribution:

0.000% <= 0.847 milliseconds (cumulative count 1)

50.000% <= 3.407 milliseconds (cumulative count 155439)

75.000% <= 3.511 milliseconds (cumulative count 228149)

87.500% <= 3.671 milliseconds (cumulative count 262828)

93.750% <= 4.791 milliseconds (cumulative count 281268)

96.875% <= 6.783 milliseconds (cumulative count 290681)

98.438% <= 6.927 milliseconds (cumulative count 295410)

99.219% <= 7.063 milliseconds (cumulative count 297758)

99.609% <= 7.175 milliseconds (cumulative count 298873)

99.805% <= 7.319 milliseconds (cumulative count 299420)

99.902% <= 7.615 milliseconds (cumulative count 299711)

99.951% <= 8.127 milliseconds (cumulative count 299857)

99.976% <= 9.335 milliseconds (cumulative count 299927)

99.988% <= 9.447 milliseconds (cumulative count 299966)

99.994% <= 9.535 milliseconds (cumulative count 299983)

99.997% <= 10.663 milliseconds (cumulative count 299991)

99.998% <= 10.703 milliseconds (cumulative count 299997)

99.999% <= 10.711 milliseconds (cumulative count 299998)

100.000% <= 10.719 milliseconds (cumulative count 299999)

100.000% <= 10.727 milliseconds (cumulative count 300000)

100.000% <= 10.727 milliseconds (cumulative count 300000)



Cumulative distribution of latencies:

0.000% <= 0.103 milliseconds (cumulative count 0)

0.003% <= 0.903 milliseconds (cumulative count 9)

0.008% <= 1.007 milliseconds (cumulative count 23)

0.012% <= 1.103 milliseconds (cumulative count 36)

0.020% <= 1.207 milliseconds (cumulative count 61)

0.029% <= 1.303 milliseconds (cumulative count 87)

0.041% <= 1.407 milliseconds (cumulative count 124)

0.054% <= 1.503 milliseconds (cumulative count 161)

0.065% <= 1.607 milliseconds (cumulative count 196)

0.082% <= 1.703 milliseconds (cumulative count 245)

0.107% <= 1.807 milliseconds (cumulative count 321)

0.135% <= 1.903 milliseconds (cumulative count 405)

0.150% <= 2.007 milliseconds (cumulative count 451)

0.164% <= 2.103 milliseconds (cumulative count 491)

4.934% <= 3.103 milliseconds (cumulative count 14802)

90.753% <= 4.103 milliseconds (cumulative count 272258)

94.011% <= 5.103 milliseconds (cumulative count 282033)

94.614% <= 6.103 milliseconds (cumulative count 283841)

99.410% <= 7.103 milliseconds (cumulative count 298231)

99.948% <= 8.103 milliseconds (cumulative count 299845)

99.968% <= 9.103 milliseconds (cumulative count 299903)

99.994% <= 10.103 milliseconds (cumulative count 299983)

100.000% <= 11.103 milliseconds (cumulative count 300000)



Summary:

  throughput summary: 238473.77 requests per second

  latency summary (msec):

          avg       min       p50       p95       p99       max

        3.608     0.840     3.407     6.375     7.015    10.727

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

  300000 requests completed in 1.26 seconds

  1000 parallel clients

  3 bytes payload

  keep alive: 1

  host configuration "save": 3600 1 300 100 60 10000

  host configuration "appendonly": no

  multi-thread: yes

  threads: 4



Latency by percentile distribution:

0.000% <= 1.295 milliseconds (cumulative count 1)

50.000% <= 3.335 milliseconds (cumulative count 151509)

75.000% <= 3.431 milliseconds (cumulative count 226229)

87.500% <= 3.607 milliseconds (cumulative count 262983)

93.750% <= 4.327 milliseconds (cumulative count 281314)

96.875% <= 6.559 milliseconds (cumulative count 290762)

98.438% <= 6.807 milliseconds (cumulative count 295466)

99.219% <= 6.903 milliseconds (cumulative count 297683)

99.609% <= 7.015 milliseconds (cumulative count 298851)

99.805% <= 7.207 milliseconds (cumulative count 299415)

99.902% <= 7.495 milliseconds (cumulative count 299709)

99.951% <= 7.975 milliseconds (cumulative count 299855)

99.976% <= 8.255 milliseconds (cumulative count 299928)

99.988% <= 8.479 milliseconds (cumulative count 299964)

99.994% <= 8.591 milliseconds (cumulative count 299982)

99.997% <= 10.959 milliseconds (cumulative count 299991)

99.998% <= 10.983 milliseconds (cumulative count 299996)

99.999% <= 11.015 milliseconds (cumulative count 299998)

100.000% <= 11.023 milliseconds (cumulative count 299999)

100.000% <= 11.031 milliseconds (cumulative count 300000)

100.000% <= 11.031 milliseconds (cumulative count 300000)



Cumulative distribution of latencies:

0.000% <= 0.103 milliseconds (cumulative count 0)

0.001% <= 1.303 milliseconds (cumulative count 2)

0.010% <= 1.407 milliseconds (cumulative count 31)

0.022% <= 1.503 milliseconds (cumulative count 66)

0.035% <= 1.607 milliseconds (cumulative count 105)

0.056% <= 1.703 milliseconds (cumulative count 168)

0.077% <= 1.807 milliseconds (cumulative count 230)

0.083% <= 1.903 milliseconds (cumulative count 250)

0.085% <= 2.007 milliseconds (cumulative count 256)

0.129% <= 2.103 milliseconds (cumulative count 388)

9.830% <= 3.103 milliseconds (cumulative count 29489)

92.365% <= 4.103 milliseconds (cumulative count 277095)

94.554% <= 5.103 milliseconds (cumulative count 283662)

95.195% <= 6.103 milliseconds (cumulative count 285584)

99.710% <= 7.103 milliseconds (cumulative count 299130)

99.961% <= 8.103 milliseconds (cumulative count 299883)

99.995% <= 9.103 milliseconds (cumulative count 299986)

100.000% <= 11.103 milliseconds (cumulative count 300000)



Summary:

  throughput summary: 238853.50 requests per second

  latency summary (msec):

          avg       min       p50       p95       p99       max

        3.513     1.288     3.335     6.015     6.871    11.031

测试结论：

华为云Flexus X实例Redis测试数据
多线程	GET	SET
未开启	171135.19	172413.80
开启	238853.50	238473.77

本次仅仅测试GET/SET的性能，没有做到全面的数据类型，但是并不妨碍总体的结果。这个效果还是非常地恐怖，出乎意料。在未开启多线程读写的情况下QPS居然达到了17W多每秒。开启了多线程读写之后，性能更是恐怖的达到了24W每秒。这个结果比最初开始了解Redis读写10W的标准整整提升了1倍多。华为云Flexus X实例Redis加速果然有点厉害，有点东西。