FIO参数

filename=/export/1.txt　支持文件系统或者裸设备，-filename=/dev/sda2或-filename=/dev/sdb

direct=1测试过程绕过机器自带的buffer，使测试结果更真实

rw=randwread 测试随机读的I/O

rw=randwrite 测试随机写的I/O

rw=randrw 测试随机混合写和读的I/O

rw=read 测试顺序读的I/O

rw=write 测试顺序写的I/O

rw=rw 测试顺序混合写和读的I/O

bs=4k 单次io的块文件大小为4k

bsrange=512-2048同上，提定数据块的大小范围

size=5g 本次的测试文件大小为5g，以每次4k的io进行测试

numjobs=30本次的测试线程为30

runtime=1000测试时间为1000秒，如果不写则一直将5g文件分4k每次写完为止

ioengine=psync io引擎使用pync方式，如果要使用libaio引擎，需要yum install libaio-devel包

rwmixwrite=30 在混合读写的模式下，写占30%group_reporting 关于显示结果的，汇总每个进程的信息

lockmem=1g 只使用1g内存进行测试

zero_buffers 用0初始化系统buffer

nrfiles=8 每个进程生成文件的数量

ioengine参数

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.

测试结果解读

4k随机读写，70%读+30%写，30个并发

fio -filename=/export/1.txt -direct=1 -iodepth 1 -thread -rw=randrw -rwmixread=70 -ioengine=psync -bs=4k -size=10G -numjobs=30 -runtime=60 -group_reporting -name=randrw_70read_4k

输出结果：

Jobs: 30 (f=30): [mmmmmmmmmmmmmmmmmmmmmmmmmmmmmm] [100.0% done] [99648K/42236K/0K /s] [24.1K/10.6K/0iops] [eta 00m:00s]

randrw_70read_4k: (groupid=0, jobs=30): err= 0: pid=4183: Sun May 5 14:56:16 2019read : io=5873.2MB, bw=100232KB/s, iops=25057 , runt=60002msec

clat (usec): min=129 , max=17247 , avg=872.95, stdev=884.86lat (usec): min=129 , max=17247 , avg=873.04, stdev=884.86clat percentiles (usec):| 1.00th=[ 141], 5.00th=[ 149], 10.00th=[ 157], 20.00th=[ 175],| 30.00th=[ 193], 40.00th=[ 788], 50.00th=[ 988], 60.00th=[ 1096],| 70.00th=[ 1192], 80.00th=[ 1304], 90.00th=[ 1448], 95.00th=[ 1576],| 99.00th=[ 2640], 99.50th=[ 8512], 99.90th=[ 9152], 99.95th=[ 9280],| 99.99th=[10048]

bw (KB/s) : min= 2680, max= 3992, per=3.34%, avg=3342.81, stdev=180.17write: io=2519.1MB, bw=43005KB/s, iops=10751 , runt=60002msec

clat (usec): min=45 , max=21382 , avg=750.66, stdev=874.94lat (usec): min=45 , max=21382 , avg=750.85, stdev=874.94clat percentiles (usec):| 1.00th=[ 52], 5.00th=[ 57], 10.00th=[ 60], 20.00th=[ 68],| 30.00th=[ 79], 40.00th=[ 644], 50.00th=[ 868], 60.00th=[ 988],| 70.00th=[ 1080], 80.00th=[ 1192], 90.00th=[ 1320], 95.00th=[ 1448],| 99.00th=[ 1816], 99.50th=[ 8384], 99.90th=[ 9024], 99.95th=[ 9280],| 99.99th=[ 9920]

bw (KB/s) : min= 1016, max= 2064, per=3.34%, avg=1434.25, stdev=131.77lat (usec) :50=0.06%, 100=11.02%, 250=26.21%, 500=0.67%, 750=2.40%lat (usec) :1000=13.91%lat (msec) :2=44.51%, 4=0.41%, 10=0.79%, 20=0.01%, 50=0.01%cpu : usr=0.19%, sys=2.19%, ctx=4148485, majf=0, minf=18446744073708663675IO depths :1=100.0%, 2=0.0%, 4=0.0%, 8=0.0%, 16=0.0%, 32=0.0%, >=64=0.0%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=0.0%issued : total=r=1503524/w=645100/d=0, short=r=0/w=0/d=0Run status group0(all jobs):

READ: io=5873.2MB, aggrb=100231KB/s, minb=100231KB/s, maxb=100231KB/s, mint=60002msec, maxt=60002msec

WRITE: io=2519.1MB, aggrb=43005KB/s, minb=43005KB/s, maxb=43005KB/s, mint=60002msec, maxt=60002msec

Disk stats (read/write):

sda: ios=1499788/643481, merge=0/1, ticks=271811/40234, in_queue=311058, util=98.88%

缩写解释：

io=执行了多少M的IO

bw=平均IO带宽

iops=IOPS

runt=线程运行时间

slat=提交延迟

clat=完成延迟

lat=响应时间

bw=带宽

cpu=利用率

IO depths=io队列

IO submit=单个IO提交要提交的IO数

IO complete=Like the above submit number, but forcompletions instead.

IO issued=The number of read/write requests issued, and how many of them were short.

IO latencies=IO完延迟的分布

io=总共执行了多少size的IO

aggrb=group总带宽

minb=最小.平均带宽.

maxb=最大平均带宽.

mint=group中线程的最短运行时间.

maxt=group中线程的最长运行时间.

ios=所有group总共执行的IO数.

merge=总共发生的IO合并数.

ticks=Number of ticks we kept the disk busy.

io_queue=花费在队列上的总共时间.

util=磁盘利用率

在8块800GB的SSD做RAID 10的存储上使用psync引擎进行测试，测试结果：

4k+70%随机读+30%随机写: 24.1K+10.6K

4k+100%随机读: 127K

4k+100%随机写: 16.2K

1k+70%随机读+30%随机写: 23.4K+10.4K

1k+100%随机读: 131K

1k+100%随机写: 13.8K

上面跑的时间较短，测试误差较大！

抄自：

https://www.cnblogs.com/raykuan/p/6914748.html

http://blog.yufeng.info/archives/1497

http://blog.yufeng.info/archives/677