Fio IO性能测试工具介绍
原创文章，转载请注明： 转载自系统技术非业余研究
本文链接地址: Fio IO性能测试工具介绍
官网：http://freshmeat.net/projects/fio/
fio is an I/O tool meant to be used both for benchmark and stress/hardware verification. It has support for 13 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. It supports Linux, FreeBSD, NetBSD, OS X, and OpenSolaris.
Ubuntu下可以用apt-get install fio安装就好。
这个工具最大的特点是使用简单，支持的文件操作非常多, 可以覆盖到我们能见到的文件使用方式:
sync：Basic read(2) or write(2) I/O. fseek(2) is used to position the I/O location.
psync：Basic pread(2) or pwrite(2) I/O.
vsync: Basic readv(2) or writev(2) I/O. Will emulate queuing by coalescing adjacents IOs into a single submission.
libaio: Linux native asynchronous I/O.
posixaio: glibc POSIX asynchronous I/O using aio_read(3) and aio_write(3).
mmap: File is memory mapped with mmap(2) and data copied using memcpy(3).
splice： splice(2) is used to transfer the data and vmsplice(2) to transfer data from user-space to the kernel.
syslet-rw： Use the syslet system calls to make regular read/write asynchronous.
sg：SCSI generic sg v3 I/O.
net ： Transfer over the network. filename must be set appropriately to `host/port’ regardless of data direction. If receiving,
only the port argument is used.
netsplice： Like net, but uses splice(2) and vmsplice(2) to map data and send/receive.
guasi The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface approach to asycnronous I/O.
还可以控制io depth对于测试磁盘的性能很有帮助，对结果的解读也做的很明白。
典型的使用如下：
fio –filename=/dev/sdc1 –direct=1 –rw=randread –bs=4k –size=60G –numjobs=64 –runtime=10 –group_reporting –name=fileXXX
玩的开心。
Post Footer automatically generated by wp-posturl plugin for wordpress.
霸爷的这篇文章bing上搜索fio+linux竟然排第一位，必须顶一下。
霸爷，不好意思啊，耽误你点时间，麻烦看看这四组数据
CMD是 fio -filename=/dev/xvdb -direct=1 -iodepth 1 -thread -rw=randrw -rwmixread=70 -ioengine=psync -bs=4k -size=50G -numjobs=32 -runtime=3600 -group_reporting -name=mytest -ioscheduler=noop
读写频率为7:3,我让它跑了一个小时，按照那边服务器商的说法是没有任何其他负载了，你看第2组的io比例，和7:3差的很远，就它和别人不一样，这个正常么，还有第3组的iops是不是相当不给力啊？另外我设-direct=0，iops显著提升，但是会周期性的突然急剧下降，这是文件缓存在刷脏页的问题吧？
打搅请见谅 Orz
打搅请见谅 Orz
CentOS 6.3 64位
read : io=1146.3MB, bw=333747 B/s, iops=81 , runt=3601413msec
write: io=501996KB, bw=142733 B/s, iops=34 , runt=3601413msec
CentOS release 6.2
read : io=1272.3MB, bw=1526.1KB/s, iops=381 , runt=3600077msec
write: io=2302.5MB, bw=670614 B/s, iops=163 , runt=3600077msec
CentOS release 5.6.
read : io=271732KB, bw=77261 B/s, iops=18 , runt=3601460msec
write: io=116056KB, bw=32998 B/s, iops=8 , runt=3601460msec
CentOS release 5.8
read : io=2155.8MB, bw=627661 B/s, iops=153 , runt=3600278msec
write: io=950684KB, bw=270395 B/s, iops=66 , runt=3600278msec
Yu Feng Reply:
January 6th, 2013 at 6:08 pm
direct=0周期性的突然急剧下降是刷脏页的问题。
-iodepth 1 这个为什么要这么设置呀？
-ioscheduler=noop 这个的目的是？
baifan Reply:
January 7th, 2013 at 12:03 pm
noop是个能做申请合并的FIFO队列，但是direct=1绕过缓存，它就成为一个单纯的先进先出的队列了，
我觉得这样更有益于测试磁盘寻道随机性能，如果是CFQ的话，队列私有会影响原始申请的位置。
iodepth不是每次提交队列的深度么？设置1不让他一次提交多个
我反正是这么理解的
Yu Feng Reply:
January 7th, 2013 at 12:05 pm
iodepth是针对全局的吧
baifan Reply:
January 7th, 2013 at 12:05 pm
能透露下TB的服务器 fio测试iops有多少嘛？？ 哈哈
Yu Feng Reply:
January 7th, 2013 at 12:07 pm
iops通常10w左右吧
baifan Reply:
January 7th, 2013 at 12:35 pm
哦 这个东西默认就是1，貌似在libaio上才有讲究
呃 这次回复的这么快
Yu Feng Reply:
January 7th, 2013 at 12:07 pm
前几天休假去了
iodepth我去查查看
!- – 10W 好有钱 iodepth我去找找看 硬件懂的少，这次测试只是客串一下，我是写代码的 哈哈
不好意思，问个弱智问题：fio命令中 ，filename到底指的是什么呢？就是指定需要测试的硬盘或分区吗？
Yu Feng Reply:
January 8th, 2013 at 11:43 am
linux下一切都是文件，设备当然也是文件，文件名叫 /dev/xxx
yue Reply:
January 8th, 2013 at 2:27 pm
嗯。如果这样直接测试裸设备，会对分区或硬盘上的数据造成影响吧？本人菜鸟，上次直接对
系统分区做了写测试，结果导致系统崩掉了，不清楚这里面的原理
Yu Feng Reply:
January 8th, 2013 at 4:13 pm
搞个分区搞，哪能整个系统盘，文件系统立马乱了。
不好意思，问个fio无法实用的问题哈~
本人使用centos6.3与6.4都试了一下，按照安装步骤，加载libaio与libaio-dev，下载相应fio2.0（官网上的），make，make install。使用fio -o1 test.fio，提示:fopen job file: No such file or directory…..
也使用fio -filename=/dev/sda -direct=1 -iodepth 1 -thread -rw=randrw -rwmixread=70 -ioengine=psync -bs=4k -size=10G -numjobs=32 -runtime=100 -group_reporting -name=mytest
提示：fopen job file: No such file or directory
请问：这是什么原因导致？
感谢不吝赐教~
liu.li Reply:
June 9th, 2013 at 5:04 pm
还有就是使用fio2.0出现这样的问题：
[root@localhost fio-2.1]# fio –filename=/dev/sda1 –direct=1 –rw=randwrite –bs=4k –size=10G –numjobs=32 –runtime=10 –group_reporting –name=aa
fio: failed parsing group_reporting=–name=aa
尝试了去掉group_reporting前边的引用“–”
[root@localhost fio-2.1]# fio –filename=/dev/sda1 –direct=1 –rw=randwrite –b
s=4k –size=10G –numjobs=32 –runtime=10 group_reporting –name=aa
aa: (g=0): rw=randwrite, bs=4K-4K/4K-4K/4K-4K, ioengine=sync, iodepth=1
…
aa: (g=0): rw=randwrite, bs=4K-4K/4K-4K/4K-4K, ioengine=sync, iodepth=1
fopen job file: No such file or directory
Yu Feng Reply:
June 9th, 2013 at 6:15 pm
最好把操作写在job文件里，一不乱，二容易俺section组织，每次运行不同的section
Yu Feng Reply:
June 9th, 2013 at 6:14 pm
fio –debug=all 就知道什么原因了。
liu.li Reply:
June 13th, 2013 at 11:28 am
运行fio -debug=all后出现以下信息，但不明白什么意思，貌似相应的包我也安装了~
fio: set all debug options
parse 32147 free options
time 32147 cycles[0]=2666
time 32147 cycles[1]=2665
time 32147 cycles[2]=2666
time 32147 cycles[3]=2666
time 32147 cycles[4]=2666
time 32147 cycles[5]=2666
time 32147 cycles[6]=2666
time 32147 cycles[7]=2666
time 32147 cycles[8]=2666
time 32147 cycles[9]=2666
time 32147 cycles[10]=2666
time 32147 cycles[11]=2666
time 32147 cycles[12]=2666
time 32147 cycles[13]=2666
time 32147 cycles[14]=2666
time 32147 cycles[15]=2666
time 32147 cycles[16]=2666
time 32147 cycles[17]=2666
time 32147 cycles[18]=2666
time 32147 cycles[19]=2666
time 32147 cycles[20]=2666
time 32147 cycles[21]=2666
time 32147 cycles[22]=2666
time 32147 cycles[23]=2666
time 32147 cycles[24]=2666
time 32147 cycles[25]=2666
time 32147 cycles[26]=2666
time 32147 cycles[27]=2666
time 32147 cycles[28]=2666
time 32147 cycles[29]=2666
time 32147 cycles[30]=2666
time 32147 cycles[31]=2666
time 32147 cycles[32]=2666
time 32147 cycles[33]=2666
time 32147 cycles[34]=2666
time 32147 cycles[35]=2666
time 32147 cycles[36]=2666
time 32147 cycles[37]=2666
time 32147 cycles[38]=2666
time 32147 cycles[39]=2666
time 32147 cycles[40]=2666
time 32147 cycles[41]=2666
time 32147 cycles[42]=2666
time 32147 cycles[43]=2666
time 32147 cycles[44]=2666
time 32147 cycles[45]=2666
time 32147 cycles[46]=2666
time 32147 cycles[47]=2666
time 32147 cycles[48]=2666
time 32147 cycles[49]=2666
time 32147 avg: 2667
time 32147 mean=2666.650000, S=0.035743
time 32147 inv_cycles_per_usec=6290
io 32147 ioengine cpuio unregistered
io 32147 ioengine mmap unregistered
io 32147 ioengine sync unregistered
io 32147 ioengine psync unregistered
io 32147 ioengine vsync unregistered
io 32147 ioengine null unregistered
io 32147 ioengine net unregistered
io 32147 ioengine netsplice unregistered
io 32147 ioengine libaio unregistered
io 32147 ioengine posixaio unregistered
io 32147 ioengine falloc unregistered
io 32147 ioengine e4defrag unregistered
io 32147 ioengine splice unregistered
io 32147 ioengine sg unregistered
io 32147 ioengine binject unregistered
profile 32147 unregister profile ‘tiobench’
liu.li Reply:
June 13th, 2013 at 1:59 pm
可以了，在job文件里添加了ioengine=mmap。调用的，好像还行，但是关于这些engine，具体代表啥意思，请霸爷给明示啊~感谢~
Yu Feng Reply:
June 13th, 2013 at 4:32 pm
ioengine=str
Defines how the job issues I/O. The following types are defined:
sync
Basic read(2) or write(2) I/O. fseek(2) is used to position the I/O location.
psync
Basic pread(2) or pwrite(2) I/O.
vsync
Basic readv(2) or writev(2) I/O. Will emulate queuing by coalescing adjacents IOs into a single submission.
libaio
Linux native asynchronous I/O. This ioengine defines engine specific options.
posixaio
POSIX asynchronous I/O using aio_read(3) and aio_write(3).
solarisaio
Solaris native asynchronous I/O.
windowsaio
Windows native asynchronous I/O.
mmap
File is memory mapped with mmap(2) and data copied using memcpy(3).
splice
splice(2) is used to transfer the data and vmsplice(2) to transfer data from user-space to the kernel.
syslet-rw
Use the syslet system calls to make regular read/write asynchronous.
sg
SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if the target is an sg character device, we use read(2) and write(2) for asynchronous I/O.
null
Doesn’t transfer any data, just pretends to. Mainly used to exercise fio itself and for debugging and testing purposes.
net
Transfer over the network. The protocol to be used can be defined with the protocol parameter. Depending on the protocol, filename, hostname, port, or listen must be specified. This ioengine defines engine specific options.
netsplice
Like net, but uses splice(2) and vmsplice(2) to map data and send/receive. This ioengine defines engine specific options.
cpuio
Doesn’t transfer any data, but burns CPU cycles according to cpuload and cpucycles parameters.
guasi
The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface approach to asycnronous I/O.
See <http://www.xmailserver.org/guasi-lib.html>.
rdma
The RDMA I/O engine supports both RDMA memory semantics (RDMA_WRITE/RDMA_READ) and channel semantics (Send/Recv) for the InfiniBand, RoCE and iWARP protocols.
external
Loads an external I/O engine object file. Append the engine filename as ‘:enginepath’.
falloc
IO engine that does regular linux native fallocate callt to simulate data transfer as fio ioengine
DDIR_READ does fallocate(,mode = FALLOC_FL_KEEP_SIZE,)
DIR_WRITE does fallocate(,mode = 0)
DDIR_TRIM does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE)
e4defrag
IO engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate defragment activity request to DDIR_WRITE event
您好，fio的测试结果如何保存到文件呢？查阅文档似乎没有发现..还望指教！
Yu Feng Reply:
August 22nd, 2013 at 9:15 pm
man fio
–output=filename
Write output to filename.
tina Reply:
February 7th, 2017 at 8:18 pm
不知现在还能答复么，怎么在测试配置文件里面将测试结果保存到文件呢？