LoongArch单机Ceph Bcache加速4K随机写性能测试

2023-10-17 23:44

本文主要是介绍LoongArch单机Ceph Bcache加速4K随机写性能测试,希望对大家解决编程问题提供一定的参考价值,需要的开发者们随着小编来一起学习吧!

LoongArch单机Ceph Bcache加速4K随机写性能测试

两块HDD做OSD

[root@ceph01 ~]# fio -direct=1 -iodepth=128 -thread -rw=randwrite -ioengine=libaio -bs=4k -size=100G -numjobs=1 -runtime=600 -group_reporting -name=mytest -filename=/dev/rbd0
mytest: (g=0): rw=randwrite, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=128
fio-3.22
Starting 1 thread
Jobs: 1 (f=1): [w(1)][100.0%][w=1605KiB/s][w=401 IOPS][eta 00m:00s]
mytest: (groupid=0, jobs=1): err= 0: pid=83763: Mon Oct 16 03:44:45 2023write: IOPS=404, BW=1620KiB/s (1659kB/s)(950MiB/600262msec); 0 zone resetsslat (usec): min=3, max=116, avg= 5.76, stdev= 4.41clat (msec): min=36, max=947, avg=316.05, stdev=71.26lat (msec): min=36, max=947, avg=316.06, stdev=71.26clat percentiles (msec):|  1.00th=[  180],  5.00th=[  215], 10.00th=[  239], 20.00th=[  264],| 30.00th=[  279], 40.00th=[  296], 50.00th=[  309], 60.00th=[  326],| 70.00th=[  342], 80.00th=[  363], 90.00th=[  397], 95.00th=[  435],| 99.00th=[  542], 99.50th=[  609], 99.90th=[  793], 99.95th=[  810],| 99.99th=[  944]bw (  KiB/s): min=  232, max= 3072, per=100.00%, avg=1622.36, stdev=394.76, samples=1198iops        : min=   58, max=  768, avg=405.58, stdev=98.69, samples=1198lat (msec)   : 50=0.02%, 100=0.01%, 250=14.65%, 500=83.73%, 750=1.44%lat (msec)   : 1000=0.15%cpu          : usr=0.11%, sys=0.30%, ctx=16672, majf=0, minf=0IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.1%issued rwts: total=0,243095,0,0 short=0,0,0,0 dropped=0,0,0,0latency   : target=0, window=0, percentile=100.00%, depth=128Run status group 0 (all jobs):WRITE: bw=1620KiB/s (1659kB/s), 1620KiB/s-1620KiB/s (1659kB/s-1659kB/s), io=950MiB (996MB), run=600262-600262msecDisk stats (read/write):rbd0: ios=0/242996, merge=0/0, ticks=0/76720597, in_queue=76980152, util=100.00%

【Bcache】一块SSD加速两块HDD(OSD)

Ceph中使用bcache_4K随机写测试_1块SDD加速2块HDD

image-20231013003009484

[root@ceph01 ceph]# fio -direct=1 -iodepth=128 -thread -rw=randwrite -ioengine=libaio -bs=4k -size=100G -numjobs=1 -runtime=600 -group_reporting -name=mytest -filename=/dev/rbd0
mytest: (g=0): rw=randwrite, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=128
fio-3.22
Starting 1 thread
Jobs: 1 (f=1): [w(1)][10.7%][w=12.1MiB/s][w=3097 IOPS][eta 08m:56s]
mytest: (groupid=0, jobs=1): err= 0: pid=37245: Thu Oct 12 22:08:43 2023write: IOPS=4065, BW=15.9MiB/s (16.7MB/s)(1024MiB/64475msec); 0 zone resetsslat (usec): min=3, max=173, avg= 5.80, stdev= 3.99clat (msec): min=9, max=336, avg=31.47, stdev=21.69lat (msec): min=9, max=336, avg=31.48, stdev=21.69clat percentiles (msec):|  1.00th=[   17],  5.00th=[   19], 10.00th=[   21], 20.00th=[   24],| 30.00th=[   26], 40.00th=[   27], 50.00th=[   28], 60.00th=[   29],| 70.00th=[   31], 80.00th=[   33], 90.00th=[   37], 95.00th=[   48],| 99.00th=[  146], 99.50th=[  180], 99.90th=[  268], 99.95th=[  284],| 99.99th=[  334]bw (  KiB/s): min= 4216, max=20288, per=100.00%, avg=16284.49, stdev=3300.36, samples=128iops        : min= 1054, max= 5072, avg=4071.00, stdev=825.06, samples=128lat (msec)   : 10=0.01%, 20=8.54%, 50=87.00%, 100=2.20%, 250=2.13%lat (msec)   : 500=0.14%cpu          : usr=0.95%, sys=3.21%, ctx=28915, majf=0, minf=0IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.1%issued rwts: total=0,262144,0,0 short=0,0,0,0 dropped=0,0,0,0latency   : target=0, window=0, percentile=100.00%, depth=128Run status group 0 (all jobs):WRITE: bw=15.9MiB/s (16.7MB/s), 15.9MiB/s-15.9MiB/s (16.7MB/s-16.7MB/s), io=1024MiB (1074MB), run=64475-64475msecDisk stats (read/write):rbd0: ios=0/261860, merge=0/0, ticks=0/8185334, in_queue=8198080, util=100.00%

【Bcache】两块SSD加速两块HDD(OSD)

image-20231014231026800

[root@ceph01 ceph]# fio -direct=1 -iodepth=128 -thread -rw=randwrite -ioengine=libaio -bs=4k -size=100G -numjobs=1 -runtime=600 -group_reporting -name=mytest -filename=/dev/rbd0
mytest: (g=0): rw=randwrite, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=128
fio-3.22
Starting 1 thread
Jobs: 1 (f=1): [w(1)][10.5%][w=15.8MiB/s][w=4047 IOPS][eta 08m:57s]
mytest: (groupid=0, jobs=1): err= 0: pid=363441: Sat Oct 14 11:19:49 2023write: IOPS=4158, BW=16.2MiB/s (17.0MB/s)(1024MiB/63037msec); 0 zone resetsslat (usec): min=3, max=624, avg= 5.76, stdev= 4.10clat (msec): min=10, max=359, avg=30.77, stdev=18.74lat (msec): min=10, max=359, avg=30.78, stdev=18.74clat percentiles (msec):|  1.00th=[   17],  5.00th=[   19], 10.00th=[   21], 20.00th=[   23],| 30.00th=[   25], 40.00th=[   27], 50.00th=[   28], 60.00th=[   29],| 70.00th=[   31], 80.00th=[   33], 90.00th=[   38], 95.00th=[   51],| 99.00th=[  111], 99.50th=[  153], 99.90th=[  262], 99.95th=[  266],| 99.99th=[  359]bw (  KiB/s): min= 3768, max=21016, per=100.00%, avg=16674.80, stdev=3007.06, samples=125iops        : min=  942, max= 5254, avg=4168.57, stdev=751.74, samples=125lat (msec)   : 20=9.54%, 50=85.45%, 100=3.49%, 250=1.38%, 500=0.13%cpu          : usr=0.99%, sys=3.24%, ctx=28037, majf=0, minf=0IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.1%issued rwts: total=0,262144,0,0 short=0,0,0,0 dropped=0,0,0,0latency   : target=0, window=0, percentile=100.00%, depth=128Run status group 0 (all jobs):WRITE: bw=16.2MiB/s (17.0MB/s), 16.2MiB/s-16.2MiB/s (17.0MB/s-17.0MB/s), io=1024MiB (1074MB), run=63037-63037msecDisk stats (read/write):rbd0: ios=0/261835, merge=0/0, ticks=0/7989253, in_queue=7996052, util=100.00%

【Bcache】两块SSD加速两块HDD(OSD)+两块SSD加速block.db和block.wal

[root@ceph01 ~]# fio -direct=1 -iodepth=128 -thread -rw=randwrite -ioengine=libaio -bs=4k -size=100G -numjobs=1 -runtime=600 -group_reporting -name=mytest -filename=/dev/rbd0
mytest: (g=0): rw=randwrite, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=128
fio-3.22
Starting 1 thread
Jobs: 1 (f=1): [w(1)][10.5%][w=16.3MiB/s][w=4162 IOPS][eta 08m:57s]
mytest: (groupid=0, jobs=1): err= 0: pid=73003: Mon Oct 16 02:34:47 2023write: IOPS=4124, BW=16.1MiB/s (16.9MB/s)(1024MiB/63559msec); 0 zone resetsslat (usec): min=3, max=109, avg= 5.69, stdev= 3.75clat (msec): min=10, max=294, avg=31.03, stdev=17.27lat (msec): min=10, max=294, avg=31.03, stdev=17.27clat percentiles (msec):|  1.00th=[   17],  5.00th=[   19], 10.00th=[   21], 20.00th=[   23],| 30.00th=[   25], 40.00th=[   27], 50.00th=[   28], 60.00th=[   30],| 70.00th=[   32], 80.00th=[   34], 90.00th=[   40], 95.00th=[   52],| 99.00th=[  110], 99.50th=[  136], 99.90th=[  226], 99.95th=[  249],| 99.99th=[  284]bw (  KiB/s): min= 6200, max=20376, per=100.00%, avg=16508.00, stdev=2659.19, samples=126iops        : min= 1550, max= 5094, avg=4126.88, stdev=664.77, samples=126lat (msec)   : 20=9.65%, 50=85.11%, 100=3.87%, 250=1.33%, 500=0.04%cpu          : usr=1.00%, sys=3.13%, ctx=25141, majf=0, minf=0IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.1%issued rwts: total=0,262144,0,0 short=0,0,0,0 dropped=0,0,0,0latency   : target=0, window=0, percentile=100.00%, depth=128Run status group 0 (all jobs):WRITE: bw=16.1MiB/s (16.9MB/s), 16.1MiB/s-16.1MiB/s (16.9MB/s-16.9MB/s), io=1024MiB (1074MB), run=63559-63559msecDisk stats (read/write):rbd0: ios=0/261407, merge=0/0, ticks=0/8062837, in_queue=8075472, util=100.00%

两块SSD做OSD

image-20231013225413334

[root@ceph01 ~]# fio -direct=1 -iodepth=128 -thread -rw=randwrite -ioengine=libaio -bs=4k -size=100G -numjobs=1 -runtime=600 -group_reporting -name=mytest -filename=/dev/rbd0
mytest: (g=0): rw=randwrite, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=128
fio-3.22
Starting 1 thread
Jobs: 1 (f=1): [w(1)][4.3%][w=34.7MiB/s][w=8883 IOPS][eta 09m:35s] 
mytest: (groupid=0, jobs=1): err= 0: pid=125901: Fri Oct 13 11:06:55 2023write: IOPS=10.2k, BW=39.8MiB/s (41.7MB/s)(1024MiB/25751msec); 0 zone resetsslat (nsec): min=3310, max=78980, avg=5364.31, stdev=3425.80clat (usec): min=2965, max=33393, avg=12565.10, stdev=3428.92lat (usec): min=2970, max=33400, avg=12570.90, stdev=3428.60clat percentiles (usec):|  1.00th=[ 6652],  5.00th=[ 7963], 10.00th=[ 8717], 20.00th=[ 9765],| 30.00th=[10552], 40.00th=[11207], 50.00th=[11994], 60.00th=[12780],| 70.00th=[13829], 80.00th=[15139], 90.00th=[17171], 95.00th=[19006],| 99.00th=[22676], 99.50th=[24511], 99.90th=[27657], 99.95th=[28705],| 99.99th=[31589]bw (  KiB/s): min=33628, max=44247, per=99.99%, avg=40717.55, stdev=2014.23, samples=51iops        : min= 8407, max=11061, avg=10179.18, stdev=503.57, samples=51lat (msec)   : 4=0.01%, 10=23.01%, 20=73.64%, 50=3.35%cpu          : usr=2.23%, sys=7.41%, ctx=20485, majf=0, minf=0IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.1%issued rwts: total=0,262144,0,0 short=0,0,0,0 dropped=0,0,0,0latency   : target=0, window=0, percentile=100.00%, depth=128Run status group 0 (all jobs):WRITE: bw=39.8MiB/s (41.7MB/s), 39.8MiB/s-39.8MiB/s (41.7MB/s-41.7MB/s), io=1024MiB (1074MB), run=25751-25751msecDisk stats (read/write):rbd0: ios=0/260926, merge=0/0, ticks=0/3220267, in_queue=3224032, util=99.89%

结论

测试环境为单机双副本!!

测试IOPSBWlat
两块HDD做OSD4041659kB/s316.06ms
【Bcache】一块SSD加速两块HDD(OSD)406516.7MB/s31.48ms
【Bcache】两块SSD加速两块HDD(OSD)415817.0MB/s30.78ms
【Bcache】两块SSD加速两块HDD(OSD)+两块SSD加速block.db和block.wal412416.9MB/s31.03ms
两块SSD做OSD10.2k41.7MB/s12.57ms
对比数据1640567MB/s7.80ms

这篇关于LoongArch单机Ceph Bcache加速4K随机写性能测试的文章就介绍到这儿,希望我们推荐的文章对编程师们有所帮助!



http://www.chinasem.cn/article/228739

相关文章

Vue3 的 shallowRef 和 shallowReactive:优化性能

大家对 Vue3 的 ref 和 reactive 都很熟悉,那么对 shallowRef 和 shallowReactive 是否了解呢? 在编程和数据结构中,“shallow”(浅层)通常指对数据结构的最外层进行操作,而不递归地处理其内部或嵌套的数据。这种处理方式关注的是数据结构的第一层属性或元素,而忽略更深层次的嵌套内容。 1. 浅层与深层的对比 1.1 浅层(Shallow) 定义

性能测试介绍

性能测试是一种测试方法,旨在评估系统、应用程序或组件在现实场景中的性能表现和可靠性。它通常用于衡量系统在不同负载条件下的响应时间、吞吐量、资源利用率、稳定性和可扩展性等关键指标。 为什么要进行性能测试 通过性能测试,可以确定系统是否能够满足预期的性能要求,找出性能瓶颈和潜在的问题,并进行优化和调整。 发现性能瓶颈:性能测试可以帮助发现系统的性能瓶颈,即系统在高负载或高并发情况下可能出现的问题

字节面试 | 如何测试RocketMQ、RocketMQ?

字节面试:RocketMQ是怎么测试的呢? 答: 首先保证消息的消费正确、设计逆向用例,在验证消息内容为空等情况时的消费正确性; 推送大批量MQ,通过Admin控制台查看MQ消费的情况,是否出现消费假死、TPS是否正常等等问题。(上述都是临场发挥,但是RocketMQ真正的测试点,还真的需要探讨) 01 先了解RocketMQ 作为测试也是要简单了解RocketMQ。简单来说,就是一个分

性能分析之MySQL索引实战案例

文章目录 一、前言二、准备三、MySQL索引优化四、MySQL 索引知识回顾五、总结 一、前言 在上一讲性能工具之 JProfiler 简单登录案例分析实战中已经发现SQL没有建立索引问题,本文将一起从代码层去分析为什么没有建立索引? 开源ERP项目地址:https://gitee.com/jishenghua/JSH_ERP 二、准备 打开IDEA找到登录请求资源路径位置

【测试】输入正确用户名和密码,点击登录没有响应的可能性原因

目录 一、前端问题 1. 界面交互问题 2. 输入数据校验问题 二、网络问题 1. 网络连接中断 2. 代理设置问题 三、后端问题 1. 服务器故障 2. 数据库问题 3. 权限问题: 四、其他问题 1. 缓存问题 2. 第三方服务问题 3. 配置问题 一、前端问题 1. 界面交互问题 登录按钮的点击事件未正确绑定,导致点击后无法触发登录操作。 页面可能存在

黑神话,XSKY 星飞全闪单卷性能突破310万

当下,云计算仍然是企业主要的基础架构,随着关键业务的逐步虚拟化和云化,对于块存储的性能要求也日益提高。企业对于低延迟、高稳定性的存储解决方案的需求日益迫切。为了满足这些日益增长的 IO 密集型应用场景,众多云服务提供商正在不断推陈出新,推出具有更低时延和更高 IOPS 性能的云硬盘产品。 8 月 22 日 2024 DTCC 大会上(第十五届中国数据库技术大会),XSKY星辰天合正式公布了基于星

业务中14个需要进行A/B测试的时刻[信息图]

在本指南中,我们将全面了解有关 A/B测试 的所有内容。 我们将介绍不同类型的A/B测试,如何有效地规划和启动测试,如何评估测试是否成功,您应该关注哪些指标,多年来我们发现的常见错误等等。 什么是A/B测试? A/B测试(有时称为“分割测试”)是一种实验类型,其中您创建两种或多种内容变体——如登录页面、电子邮件或广告——并将它们显示给不同的受众群体,以查看哪一种效果最好。 本质上,A/B测

衡石分析平台使用手册-单机安装及启动

单机安装及启动​ 本文讲述如何在单机环境下进行 HENGSHI SENSE 安装的操作过程。 在安装前请确认网络环境,如果是隔离环境,无法连接互联网时,请先按照 离线环境安装依赖的指导进行依赖包的安装,然后按照本文的指导继续操作。如果网络环境可以连接互联网,请直接按照本文的指导进行安装。 准备工作​ 请参考安装环境文档准备安装环境。 配置用户与安装目录。 在操作前请检查您是否有 sud

从状态管理到性能优化:全面解析 Android Compose

文章目录 引言一、Android Compose基本概念1.1 什么是Android Compose?1.2 Compose的优势1.3 如何在项目中使用Compose 二、Compose中的状态管理2.1 状态管理的重要性2.2 Compose中的状态和数据流2.3 使用State和MutableState处理状态2.4 通过ViewModel进行状态管理 三、Compose中的列表和滚动

PR曲线——一个更敏感的性能评估工具

在不均衡数据集的情况下,精确率-召回率(Precision-Recall, PR)曲线是一种非常有用的工具,因为它提供了比传统的ROC曲线更准确的性能评估。以下是PR曲线在不均衡数据情况下的一些作用: 关注少数类:在不均衡数据集中,少数类的样本数量远少于多数类。PR曲线通过关注少数类(通常是正类)的性能来弥补这一点,因为它直接评估模型在识别正类方面的能力。 精确率与召回率的平衡:精确率(Pr