I/O性能测试---fio

                   I/O性能测试---fio

 

官方网站：

http://freecode.com/projects/fio

http://brick.kernel.dk/snaps/

 

简介：

fio is an I/O tool meant to be used both for benchmark and stress/hardware verification. It has support for 19 different types of I/O engines (sync, mmap, libaio, posixaio, SG v3, splice, null, network, syslet, guasi, solarisaio, and more), I/O priorities (for newer Linux kernels), rate I/O, forked or threaded jobs, and much more. It can work on block devices as well as files. fio accepts job descriptions in a simple-to-understand text format. Several example job files are included. fio displays all sorts of I/O performance information, including complete IO latencies and percentiles. Fio is in wide use in many places, for both benchmarking, QA, and verification purposes. It supports Linux, FreeBSD, NetBSD, OpenBSD, OS X, OpenSolaris, AIX, HP-UX, Android, and Windows.

 

安装(以CentOS6为例)：
1.yum安装(确保有epel源)
yum –y install fio
2.源码安装
yum –y install gcc gcc-c++ make libaio-devel
wget http://brick.kernel.dk/snaps/fio-2.2.6.tar.bz2
tar –xvf fio-2.2.6.tar.bz2 –C /usr/local/src
cd /usr/local/src/fio-2.2.6
make && make install

测试方法：
1.顺序读(read)
fio -filename=/dev/vdb -direct=1 -iodepth 128 -thread -rw=read -ioengine=libaio -bs=512K -numjobs=1 -runtime=120 -group_reporting -name=test

SSD iodepth用128，普通云主机iodepth用64，numjobs用1就可以，否则测出来数据会偏大，bs需要512K以上

  2.顺序写(write)
fio -filename=/dev/vdb -direct=1 -iodepth 128 -thread -rw=write -ioengine=libaio -bs=512K -numjobs=4 -runtime=120 -group_reporting -name=test

SSD iodepth用128，普通云主机iodepth用64，numjobs不用太大，bs需512K以上

  3.随机读(randread)
fio -filename=/dev/vdb -direct=1 -iodepth 128 -thread -rw=randread -ioengine=libaio -bs=4K -numjobs=128 -runtime=120 -group_reporting -name=test

bs根据测试的块大小进行调整（随机主要看4K，8K），SSD numjobs为128，普通云主机用4或8就可以了

  4.随机写(randwrite)
fio -filename=/dev/vdb -direct=1 -iodepth 128 -thread -rw=randwrite -ioengine=libaio -bs=4K -numjobs=32 -runtime=120 -group_reporting -name=test

bs根据测试的块大小进行调整（随机主要看4K，8K），SSD numjobs为32，普通云主机用4或8就可以了，numjobs太大的话反而会带来副作用

  5.顺序混合读写(readwrite)
fio -filename=/dev/vdb -direct=1 -iodepth 128 -thread -rw=rw -rwmixread=100 -ioengine=libaio -bs=512K -numjobs=8 -runtime=120 -group_reporting -name=test

  6.随机混合读写(randrw)
fio -filename=/dev/vdb -direct=1 -iodepth 128 -thread -rw=randrw -rwmixread=100 -ioengine=libaio -bs=4K -numjobs=8 -runtime=120 -group_reporting -name=test
SSD iodepth用128，普通云主机用64

混合读写的比例调整-rwmixread就可以了，100表示全是读，0表示全是写，50表示一半读一半写，根据需要调整。目前混合读写最好引导用户直接测裸盘，如果测文件的话会暴露出来我们的读IO较差，且读IO过多会导致写IO性能下降。

还可以通过以下配置静默测试后通过plot绘制图表

[global]

ioengine=libaio

iodepth=128

time_based

direct=1

thread=1

group_reporting

randrepeat=0

norandommap

numjobs=32

timeout=6000

runtime=120

[randread-4k]

rw=randread

bs=4k

filename=/dev/sdg

rwmixread=100

stonewall

[randwrite-4k]

rw=randwrite

bs=4k

filename=/dev/sdg

stonewall

[read-512k]

rw=read

bs=512k

filename=/dev/sdg

stonewall

[write-512k]

rw=write

bs=512k

filename=/dev/sdg

stonewall