转自:Jarvis Wang 

FIO测试脚本

FIO是一个UNIX环境下的硬盘和阵列性能测试工具，具有丰富的参数，可以调用各种各样的读写函数进行测试，帮助文档参见：http://www.cse.unsw.edu.au/~aaronc/iosched/doc/fio.1.html

下面是我写的两个FIO测试脚本，一个用于随机I/O测试，一个用于多路顺序I/O测试：

random_io_test.sh

----------------------------------------------------------------------------

target=$1

size=`cat /proc/partitions | grep $target | awk '{print $3}'`

model=`lsscsi | grep $target$ | awk '{printf "%s_%s",$4,$5}'`

echo "CASEID,IOPS,RIOPS,RLAT,WIOPS,WLAT" | tee $model.$target.random.csv

echo '' > $model.$target.random.log

cp libaio.so.1 /lib64/

for read in 0 20 40 60 80 100

do

    for block in 4 8 16 32

    do

        for queue in 1 2 4 8 16

        do

            name=R_${read}R_${block}K_${queue}Q

            ./fio --name=$name --rw=randrw --direct=1 --norandommap --ioengine=libaio --runtime=60s --ioscheduler=noop --size=${size}k --filename=/dev/$target  --rwmixread=$read --bs=${block}k --iodepth=$queue --minimal >> $model.$target.random.log

            tail -2 $model.$target.random.log | awk -F ";" -v BLOCK=$block '/K/{printf "%s,%.2f,%.2f,%.2f,%.2f,%.2f\n",$1,($5+$21)/BLOCK,$5/BLOCK,$13/1000,$21/BLOCK,$29/1000}' | tee -a $model.$target.random.csv

        done #end queue depth

    done #end block size

done #end read percent

----------------------------------------------------------------------------

sequential_io_test.sh

----------------------------------------------------------------------------

target=$1

size=`cat /proc/partitions | grep $target | awk '{print $3}'`

model=`lsscsi | grep $target$ | awk '{printf "%s_%s",$4,$5}'`

echo "CASEID,MBPS,RMBPS,RLAT,WMBPS,WLAT" | tee $model.$target.seqential.csv

echo '' > $model.$target.seqential.log

cp libaio.so.1 /lib64/

for read in 0 20 40 60 80 100

do

    for block in 64 128 256 512

    do

        for stream in 1 2 4 8 16

        do

            name=S_${read}R_${block}K_${stream}S

            ./fio --name=$name -rw=rw --direct=1 --ioengine=libaio --runtime=60s --ioscheduler=noop --filename=/dev/$target --rwmixread=$read --bs=${block}k --iodepth=1 --zonesize=$[size/stream]k --numjobs=$stream --group_reporting --minimal >> $model.$target.seqential.log

            tail -2 $model.$target.seqential.log | awk -F ";" '/K/{printf "%s,%.2f,%.2f,%.2f,%.2f,%.2f\n",$1,($5+$21)/1024,$5/1024,$13/1000,$21/1024,$29/1000}' | tee -a $model.$target.seqential.csv

        done #end queue depth

    done #end block size

done #end read percent

----------------------------------------------------------------------------

 

Shell编程相关资料

Bash新手指南 

Linux Shell Scripting Tutorial v1.05r3

awk使用详细 

Linux Shell编程 

Linux系统用户账号的管理 

文件和目录访问权限设置