IOZONE说明

iozone是一个文件系统的benchmark工具，可以测试不同的操作系统中文件系统的读写性能。 可以测试 Read, write, re-read,re-write, read backwards, read strided, fread, fwrite, random read, pread, mmap, aio_read, aio_write 等等不同的模式下的硬盘的性能。 测试的时候请注意，设置的测试文件的大小一定要大过你的内存（最佳为内存的两倍大小），不然linux会给你的读写的内容进行缓存。会使数值非常不真实。

IOZONE安装包

iozone3_430.tar

IOZONE安装步骤

[root@linux156 bin]# tar -xvf iozone3_430.tar

[root@linux156 bin]# cd /usr/linkapp/bin/iozone3_430/src/current/

[root@linux156 current]# make linux

查找：

[root@linux156 iozone3_430]# find /-name iozone

IOZONE测试命令

[root@linux156 current]# ./iozone -zx -b iozone-test.xls -g 8g -y 64k -i 0 -i 1

说明：

-az——执行不同粒度（record size）和测试文（file size）的自动化测试，默认情况下record size从4K—16M，file size从64K—512M。

-b——指定测试后生成的excel文件。

         -g——指定最大的file size大小。（一般为内存的两倍）

         -y——指定最小的record size数值。

         -i——指定测试的种类。（0代表write/rewrite，1代表read/re-read

IOZONE执行结果

 [root@linux156 current]# ./iozone -zx -b iozone-test.xls -g 8g -y 64k -i 0 -i 1

Iozone: Performance Test of File I/O

                Version $Revision: 3.430 $

                Compiled for 64 bit mode.

                Build: linux

 

Contributors:William Norcott, Don Capps, Isom Crawford, Kirby Collins

                     Al Slater, Scott Rhine, Mike Wisner, Ken Goss

                     Steve Landherr, Brad Smith, Mark Kelly, Dr. Alain CYR,

                     Randy Dunlap, Mark Montague, Dan Million, Gavin Brebner,

                     Jean-Marc Zucconi, Jeff Blomberg, Benny Halevy, Dave Boone,

                     Erik Habbinga, Kris Strecker, Walter Wong, Joshua Root,

                     Fabrice Bacchella, Zhenghua Xue, Qin Li, Darren Sawyer,

                     Vangel Bojaxhi, Ben England, Vikentsi Lapa.

 

        Run began: Thu May  7 16:09:36 2015

 

        Cross over of record size disabled.

        Stonewall disabled

        Using maximum file size of 8388608 kilobytes.

        Using Minimum Record Size 64 kB

        Command line used: ./iozone -zx -b iozone-test.xls -g 8g -y 64k -i 0 -i 1

        Output is in kBytes/sec

        Time Resolution = 0.000001 seconds.

        Processor cache size set to 1024 kBytes.

        Processor cache line size set to 32 bytes.

        File stride size set to 17 * record size.

random    random     bkwd    record    stride   

kB  reclen    write  rewrite    read    reread    read     write     read   rewrite      read   fwrite frewrite    fread  freread

             512       4   539459  1391933  3220553  3970896                                                                         

 

iozone test complete.

Excel output is below:

 

"Writer report"

"64"  "128"  "256"  "512"  "1024"  "2048"  "4096"  "8192"  "16384"

"512"   539459

 

"Re-writer report"

"64"  "128"  "256"  "512"  "1024"  "2048"  "4096"  "8192"  "16384"

"512"   1391933

 

"Reader report"

"64"  "128"  "256"  "512"  "1024"  "2048"  "4096"  "8192"  "16384"

"512"   3220553

 

"Re-Reader report"

"64"  "128"  "256"  "512"  "1024"  "2048"  "4096"  "8192"  "16384"

"512"   3970896

 

说明：

左侧第一列：测试文件大小

左侧第二列：record size

后面几列为对应不通测试方法的测试结果（单位为kb/s）

 

FIO说明

FIO是测试IOPS的非常好的工具，用来对硬件进行压力测试和验证，支持13种不同的I/O引擎，包括:sync,mmap, libaio, posixaio, SG v3, splice, null, network, syslet, guasi, solarisaio 等等。fio 官网地址：http://freshmeat.net/projects/fio/

FIO安装包

fio-2.1.10.tar.gz

libaio-devel-0.3.107-8.fc11.s390x.rpm

FIO安装步骤

[root@linux156 bin]#rpm -ivh libaio-devel-0.3.107-8.fc11.s390x.rpm

[root@linux156 bin]# tar -zxvf fio-2.0.10.tar.gz

[root@linux156 bin]#cd fio-2.0.10.tar.gz

[root@linux156 bin]#cd fio-2.0.10

[root@linux156 bin]./configure

[root@linux156 bin]make

[root@linux156 bin]make install

FIO测试命令

1）sync - Randread

[root@linux156 bin]# fio -filename=/usr/linkapp/fio-test01 -direct=1 -rw=randread -size 5G -numjobs=8 -runtime=120 -group_reporting -name=randread_4_psync

randread_4_psync: (g=0): rw=randread, bs=4K-4K/4K-4K/4K-4K, ioengine=sync, iodepth=1

...

fio-2.1.10

Starting 8 processes

randread_4_psync: Laying out IO file(s) (1 file(s) / 5120MB)

Jobs: 8 (f=8): [rrrrrrrr] [100.0% done] [4439KB/0KB/0KB /s] [1109/0/0 iops] [eta 00m:00s]

randread_4_psync: (groupid=0, jobs=8): err= 0: pid=6070: Thu May  7 17:50:02 2015

  read : io=554892KB, bw=4623.9KB/s,iops=1155, runt=120006msec

    clat (usec): min=204, max=127599, avg=6914.62, stdev=4660.99

     lat (usec): min=205, max=127599, avg=6914.96, stdev=4660.99

    clat percentiles (usec):

     |  1.00th=[ 1400],  5.00th=[ 2192], 10.00th=[ 2832], 20.00th=[ 3664],

     | 30.00th=[ 4320], 40.00th=[ 4960], 50.00th=[ 5664], 60.00th=[ 6496],

     | 70.00th=[ 7648], 80.00th=[ 9280], 90.00th=[12480], 95.00th=[15680],

     | 99.00th=[24192], 99.50th=[28800], 99.90th=[39680], 99.95th=[43776],

     | 99.99th=[60160]

    bw (KB  /s): min=  419, max=  725, per=12.50%, avg=578.10, stdev=50.29

    lat (usec) : 250=0.02%, 500=0.05%, 750=0.03%, 1000=0.13%

    lat (msec) : 2=3.48%, 4=21.21%, 10=58.02%, 20=14.86%, 50=2.19%

    lat (msec) : 100=0.02%, 250=0.01%

  cpu          : usr=0.11%, sys=0.53%, ctx=138852, majf=0, minf=282

  IO 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=138723/w=0/d=0, short=r=0/w=0/d=0

     latency   : target=0, window=0, percentile=100.00%, depth=1

 

Run status group 0 (all jobs):

   READ: io=554892KB, aggrb=4623KB/s, minb=4623KB/s, maxb=4623KB/s, mint=120006msec, maxt=120006msec

 

Disk stats (read/write):

  sda: ios=138530/126, merge=0/81, ticks=956991/3887, in_queue=960887, util=100.00%

 

 

2）sync - randwrite

[root@linux156 bin]# fio -filename=/usr/linkapp/fio-test01 -direct=1 -rw=randwrite -size 5G -numjobs=8 -runtime=120 -group_reporting -name=randread_4_psync

randread_4_psync: (g=0): rw=randwrite, bs=4K-4K/4K-4K/4K-4K, ioengine=sync, iodepth=1

...

fio-2.1.10

Starting 8 processes

Jobs: 7 (f=7): [wwEwwwww] [0.1% done] [0KB/375KB/0KB /s] [0/93/0 iops] [eta 02d:11h:29m:59s]

randread_4_psync: (groupid=0, jobs=8): err= 0: pid=6110: Thu May  7 17:58:33 2015

  write: io=47184KB, bw=402395B/s, iops=98, runt=120072msec

    clat (msec): min=2, max=20768, avg=43.43, stdev=591.89

     lat (msec): min=2, max=20768, avg=43.43, stdev=591.89

    clat percentiles (msec):

     |  1.00th=[    7],  5.00th=[    8], 10.00th=[    8], 20.00th=[    9],

     | 30.00th=[    9], 40.00th=[   10], 50.00th=[   10], 60.00th=[   11],

     | 70.00th=[   11], 80.00th=[   12], 90.00th=[   13], 95.00th=[   17],

     | 99.00th=[   28], 99.50th=[   41], 99.90th=[11076], 99.95th=[12780],

     | 99.99th=[16712]

    bw (KB  /s): min=    0, max=  447, per=76.78%, avg=300.97, stdev=166.24

    lat (msec) : 4=0.01%, 10=60.01%, 20=37.50%, 50=2.03%, 100=0.04%

    lat (msec) : 750=0.01%, >=2000=0.40%

  cpu          : usr=0.05%, sys=0.05%, ctx=23716, majf=0, minf=273

  IO 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=0/w=11796/d=0, short=r=0/w=0/d=0

     latency   : target=0, window=0, percentile=100.00%, depth=1

 

Run status group 0 (all jobs):

  WRITE: io=47184KB, aggrb=392KB/s, minb=392KB/s, maxb=392KB/s, mint=120072msec, maxt=120072msec

 

Disk stats (read/write):

  sda: ios=0/11933, merge=0/145, ticks=0/122419, in_queue=122417, util=99.60%

 

3）Libaio – randwrite 分区

[root@linux156 bin]# fio -ioengine=libaio -bs=4k -direct=1 -thread -rw=randwrite -size=100G -filename=/dev/sda3 -name=4kb-randwrite-test -iodepth=32 -runtime=60

4kb-randwrite-test: (g=0): rw=randwrite, bs=4K-4K/4K-4K/4K-4K, ioengine=libaio, iodepth=32

fio-2.1.10

Starting 1 thread

Jobs: 1 (f=1): [w] [0.0% done] [0KB/12KB/0KB /s] [0/3/0 iops] [eta 05d:20h:27m:36s]

4kb-randwrite-test: (groupid=0, jobs=1): err= 0: pid=7944: Fri May  8 13:13:20 2015

  write: io=12772KB, bw=213837B/s, iops=52, runt= 61161msec

    slat (usec): min=6, max=169, avg=19.00, stdev= 7.54

    clat (msec): min=99, max=5254, avg=610.66, stdev=558.00

     lat (msec): min=99, max=5254, avg=610.68, stdev=558.00

    clat percentiles (msec):

     |  1.00th=[  159],  5.00th=[  219], 10.00th=[  269], 20.00th=[  322],

     | 30.00th=[  363], 40.00th=[  404], 50.00th=[  449], 60.00th=[  490],

     | 70.00th=[  562], 80.00th=[  734], 90.00th=[ 1090], 95.00th=[ 1582],

     | 99.00th=[ 3228], 99.50th=[ 3818], 99.90th=[ 5014], 99.95th=[ 5211],

     | 99.99th=[ 5276]

    bw (KB  /s): min=   18, max=  374, per=100.00%, avg=208.68, stdev=72.03

    lat (msec) : 100=0.03%, 250=7.77%, 500=53.65%, 750=19.04%, 1000=7.27%

    lat (msec) : 2000=8.89%, >=2000=3.35%

  cpu          : usr=0.00%, sys=0.24%, ctx=3176, majf=0, minf=5

  IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.3%, 16=0.5%, 32=99.0%, >=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=0.1%, 64=0.0%, >=64=0.0%

     issued    : total=r=0/w=3193/d=0, short=r=0/w=0/d=0

     latency   : target=0, window=0, percentile=100.00%, depth=32

 

Run status group 0 (all jobs):

  WRITE: io=12772KB, aggrb=208KB/s, minb=208KB/s, maxb=208KB/s, mint=61161msec, maxt=61161msec

 

Disk stats (read/write):

  sda: ios=178/3253, merge=346/45, ticks=19971/1939592, in_queue=2481531, util=100.00%

You have new mail in /var/spool/mail/root

 

4）Libaio – randread 目录

[root@linux156 bin]# fio -directory=/home/test -iodepth=64 -ioengine=libaio -direct=1 -rw=randread -bs=4k -size=1G -numjobs=4 -runtime=20 -group_reporting -name=test-rand-read3

test-rand-read3: (g=0): rw=randread, bs=4K-4K/4K-4K/4K-4K, ioengine=libaio, iodepth=64

...

fio-2.1.10

Starting 4 processes

test-rand-read3: Laying out IO file(s) (1 file(s) / 1024MB)

test-rand-read3: Laying out IO file(s) (1 file(s) / 1024MB)

test-rand-read3: Laying out IO file(s) (1 file(s) / 1024MB)

test-rand-read3: Laying out IO file(s) (1 file(s) / 1024MB)

Jobs: 3 (f=3): [rrrE] [6.2% done] [12547KB/0KB/0KB /s] [3136/0/0 iops] [eta 05m:15s] 

test-rand-read3: (groupid=0, jobs=4): err= 0: pid=8888: Fri May  8 14:09:38 2015

  read : io=266628KB, bw=13277KB/s, iops=3319, runt= 20082msec

    slat (usec): min=2, max=81244, avg=999.62, stdev=5142.03

    clat (msec): min=1, max=1022, avg=76.05, stdev=37.53

     lat (msec): min=1, max=1022, avg=77.05, stdev=37.95

    clat percentiles (msec):

     |  1.00th=[   15],  5.00th=[   25], 10.00th=[   35], 20.00th=[   46],

     | 30.00th=[   55], 40.00th=[   63], 50.00th=[   72], 60.00th=[   81],

     | 70.00th=[   90], 80.00th=[  103], 90.00th=[  123], 95.00th=[  141],

     | 99.00th=[  186], 99.50th=[  210], 99.90th=[  297], 99.95th=[  359],

     | 99.99th=[  529]

    bw (KB  /s): min= 2058, max= 4160, per=24.98%, avg=3316.90, stdev=353.39

    lat (msec) : 2=0.01%, 4=0.02%, 10=0.50%, 20=2.11%, 50=22.50%

    lat (msec) : 100=52.85%, 250=21.81%, 500=0.19%, 750=0.01%, 1000=0.01%

    lat (msec) : 2000=0.01%

  cpu          : usr=0.21%, sys=0.86%, ctx=5912, majf=0, minf=374

  IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.2%, >=64=99.6%

     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=0.1%, >=64=0.0%

     issued    : total=r=66657/w=0/d=0, short=r=0/w=0/d=0

     latency   : target=0, window=0, percentile=100.00%, depth=64

 

Run status group 0 (all jobs):

   READ: io=266628KB, aggrb=13276KB/s, minb=13276KB/s, maxb=13276KB/s, mint=20082msec, maxt=20082msec

 

Disk stats (read/write):

  sda: ios=66633/6, merge=18/3, ticks=3196241/872, in_queue=3198791, util=99.61%

 

附录说明

1）配置参数说明

filename=/dev/sdb1       测试文件名称，通常选择需要测试的盘的data目录。
direct=1                 测试过程绕过机器自带的buffer。使测试结果更真实。
rw=randwrite             测试随机写的I/O
rw=randrw                测试随机写和读的I/O
bs=16k                   单次io的块文件大小为16k
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
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。

此外
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。

 

read 顺序读

write 顺序写

rw,readwrite 顺序混合读写

 

randwrite 随机写

randread 随机读

randrw 随机混合读写

2）测试结果说明

io总的输入输出量 

bw：带宽   KB/s 

iops：每秒钟的IO数

runt：总运行时间

lat (msec)：延迟(毫秒)

msec： 毫秒

usec： 微秒