使用步骤：
1.设置一个定时任何 执行getperf.sh，采集性能数据2.将采集到性能数据文件，如：192.168.1.1.tar.gz 解压3.将性能分析的脚步performance_analyse.sh 放到解压后的目录中4.安装gnuplot程序5.将字体文件夹，拷贝到/usr/share/fonts/目录6.直接运行 bash performance_analyse.sh本文提到的脚本和字体文件可以在下载中心下载，链接如下：
getperf.sh脚本内容如下：
##################################################说明：# 定义时间TIMES参数，表示采集的次数，# 采集时间 = TIMES * INTERVAL#eg:# TIMES=180 , INTERVAL=10 采集时间就是半小时###################################################!/bin/bashTIMES=3600INTERVAL=2PWD=`pwd`TIME=`date "+%F %H:%M:%S"`TAR=`whereis tar|awk -F ":" '{print $2}'|awk '{print $1}'`SAR=`whereis sar|awk -F ":" '{print $2}'|awk '{print $1}'`IOSTAT=`whereis iostat|awk -F ":" '{print $2}'|awk '{print $1}'`# Check Moniter ToolSysInfo(){ echo "sysip : $SYSIP"|tee $PWD/$SYSIP/sysinfo echo "starttime : $TIME" |tee -a $PWD/$SYSIP/sysinfo /sbin/ifconfig >>$PWD/$SYSIP/sysinfo echo "===================================" >>$PWD/$SYSIP/sysinfo /usr/sbin/dmidecode >>$PWD/$SYSIP/sysinfo echo "===================================" >>$PWD/$SYSIP/sysinfo /bin/cat /proc/cpuinfo >> $PWD/$SYSIP/sysinfo echo "===================================" >>$PWD/$SYSIP/sysinfo /sbin/fdisk -l >> $PWD/$SYSIP/sysinfo echo "===================================" >>$PWD/$SYSIP/sysinfo /bin/df -Th >>$PWD/$SYSIP/sysinfo echo "===================================" >>$PWD/$SYSIP/sysinfo /usr/bin/free -m >> $PWD/$SYSIP/sysinfo echo "===================================" >>$PWD/$SYSIP/sysinfo echo ""}CheckEnv(){ PUB_IP=`/sbin/ifconfig |grep "inet addr" | awk -F: '{print $2}'| awk '{print $1}'|grep -v "172\.\|10\.\|127\.\|192\."|sed -n 1p` PRI_IP=`/sbin/ifconfig |grep "inet addr" | awk -F: '{print $2}'| awk '{print $1}'|grep "10\.\|127\.\|192\."|sed -n 1p` if [ "snda$PUB_IP" == "snda" ];then SYSIP=$PRI_IP else SYSIP=$PUB_IP fi if [ -d $PWD/$SYSIP ];then rm -rf $PWD/$SYSIP fi mkdir -p $PWD/$SYSIP if ! grep iostat /usr/bin/iostat ;then yum -y install sysstat fi}GetPerf(){ CPUUSAGE="$PWD/$SYSIP/cpuusage.log" MEMUSAGE="$PWD/$SYSIP/memusage.log" DISKUSAGE="$PWD/$SYSIP/diskusage.log" NETWORK="$PWD/$SYSIP/network.log" $SAR -P ALL $INTERVAL $TIMES>> $CPUUSAGE & $IOSTAT -dkx $INTERVAL $TIMES>> $DISKUSAGE & $SAR -n DEV $INTERVAL $TIMES>> $NETWORK & $SAR -r $INTERVAL $TIMES>> $MEMUSAGE & for ((i=0;i<$TIMES;i++)) do sleep $INTERVAL done}CheckEnvSysInfoGetPerf#在同一台机器上第二次采集数据时，会删除之前采集的数据，重新采集#采集完成之后，会生产一个以 IP.tar.gz的压缩包。将这个压缩包，放到#分析脚本performance_analyse.sh 的同级目录。if [ -d $PWD/$SYSIP ];then cd $PWD rm -f $SYSIP.tar.gz tar zcvf $SYSIP.tar.gz $SYSIPfi performance_analyse.sh脚本内容如下：
################################################### 这个脚本的作用是处理由性能采集脚本收集到的性能数据#然后使用gunplot生产直观的性能图。###################################################!/bin/bashSysInfo(){ local file=$1 local productname="unknow" local cpumodel="unknow" local cpucore="unknow" local cpumhz="unknow" local physical="unknow" local realcpucore="unknow" local diskpart="unknow" local memory="unknow" ipaddr=`cat $file |grep -i "net addr"|grep -v "127"` productname=`cat $file |grep -i "product name"` cpumodel=`cat $file |grep -i "model name"|uniq -d` cpucore=`cat $file |grep "processor"|wc -l` cpumhz=`cat $file |grep -i "cpu MHz"|uniq -d` physical=`cat $file |grep -i "physical id"|sort -n|uniq -d|wc -l` realcpucore=`cat $file |grep -i "cpu cores"|uniq -d|awk -F ":" '{print $2}'` memory=` cat $file |grep -i -EB1 "mem:"` diskpart=`cat $file |grep -i "disk"|grep -E "[shv][d][a-z]"` echo "System Information:"|tee -a $REPORTFILE echo "IP address:"|tee -a $REPORTFILE echo -e "$ipaddr"|tee -a $REPORTFILE echo "$productname" |tee -a $REPORTFILE echo -e "\t$cpumodel" |tee -a $REPORTFILE echo -e "\tCPU cores\t:$cpucore"|tee -a $REPORTFILE echo -e "\t$cpumhz"|tee -a $REPORTFILE echo -e "\tPhysical cpu number:$physical"|tee -a $REPORTFILE echo -e "\tEach CPU real core:$realcpucore"|tee -a $REPORTFILE echo "$diskpart"|tee -a $REPORTFILE echo -e "Memory(MB):\n$memory"|tee -a $REPORTFILE}CpuAllUsage(){ local file=$1 cat $file|grep -i "all"|grep -v -i -E "average|linux|system" >$GNUPLOTFOLDER/sar_cpu.$$ TITLE=`cat $file |sed "/^$/d"|grep -v -i "average|linux"|sed 1d|sed -n 1p` local SOURCE_SAR_CPU="$GNUPLOTFOLDER/sar_cpu.$$" local USER_UASGE=`echo $TITLE |awk '{print $3}'` local NICE_UASGE=`echo $TITLE |awk '{print $4}'` local SYSTEM_UASGE=`echo $TITLE |awk '{print $5}'` local IOWAIT_UASGE=`echo $TITLE |awk '{print $6}'` local STEAL_UASGE=`echo $TITLE |awk '{print $7}'` local IDLE_UASGE=`echo $TITLE |awk '{print $8}'` local cpuusagemax=`cat $SOURCE_SAR_CPU|awk '{print $3+$4+$5+$6+$7}'|sort -r|sed -n 1p` local Tmp_ylable=`echo $cpuusagemax|awk -F "." '{print $1}'` local ylable=`echo $Tmp_ylable+5|bc` local cpuusagemin=`cat $SOURCE_SAR_CPU|awk '{print $3+$4+$5+$6+$7}'|sort|sed -n 1p` local cpuusageavg=`awk 'BEGIN{total=0}{total+=$8}END{print 100-total/NR}' $SOURCE_SAR_CPU` echo "`date '+%F %H:%M:%S'`: CPU Performance analysis" |tee -a $REPORTFILE echo -e "\t1.System Cpu load(%) \tmax=$cpuusagemax,average=$cpuusageavg,mim=$cpuusagemin" |tee -a $REPORTFILE/usr/local/bin/gnuplot --persist < $GNUPLOTFOLDER/idle_sum.$$ for (( i=0;i<=$corenu;i++ )) do cat $file |grep -v -i -E "average|system|all|linux"|sed "/^$/d"|awk "(\$2==$i){print}"|awk '{print $1 ," ",100-$8}'>$GNUPLOTFOLDER/$i.txt local idlesum=`awk 'BEGIN{total=0}{total+=$2}END{print total}' $GNUPLOTFOLDER/$i.txt` echo $i $idlesum >>$GNUPLOTFOLDER/idle_sum.$$ done first_load=`cat $GNUPLOTFOLDER/idle_sum.$$|sort -n -k 2 -r|sed -n 1p|awk '{print $1}'` second_load=`cat $GNUPLOTFOLDER/idle_sum.$$|sort -n -k 2 -r|sed -n 2p|awk '{print $1}'` third_load=`cat $GNUPLOTFOLDER/idle_sum.$$|sort -n -k 2 -r|sed -n 3p|awk '{print $1}'` load=($first_load $second_load $third_load) echo -e "\t2.Each core load:" local cpuload=("First" "Second" "Third") local nu=0 for i in ${load[@]} do local coreloadmax=`cat $GNUPLOTFOLDER/$i.txt|sort -n -k 2 -r|sed -n 1p|awk '{print $2}'` local coreloadavg=`awk 'BEGIN{total=0}{total+=$2}END{print total/NR}' $GNUPLOTFOLDER/$i.txt` local coreloadmin=`cat $GNUPLOTFOLDER/$i.txt|sort -n -k 2|sed -n 1p|awk '{print $2}'` echo -e "\t\t\t Load ${cpuload[$nu]} core $i : max=$coreloadmax , avg=$coreloadavg , min=$coreloadmin"|tee -a $REPORTFILE nu=`echo $nu+1|bc` done for ((i=0;i<=corenu;i++)) do if [ $i -eq $first_load ];then LW=4 elif [ $i -eq $second_load ];then LW=3 elif [ $i -eq $third_load ];then LW=2 else LW=1 fi TMP1="$GNUPLOTFOLDER/$i.txt" TMP2="using 1:2 with l lw $LW" TMP3="core $i " PLOT="$PLOT \"$TMP1\" $TMP2 title \"$TMP3\"," done local tmp_ylabel_range=`cat $file|grep -v -i -E "average|system|all|linux"|sed "/^$/d"|awk '{print 100-$8}'|sort -n -r|sed -n 1p|awk -F "." '{print $1}'|sed -n 1p` local ylabel_range=`echo $tmp_ylabel_range+5|bc`/usr/local/bin/gnuplot --persist < $GNUPLOTFOLDER/memory.$$ SOURCE_FILE=$GNUPLOTFOLDER/memory.$$ local memtotal=`awk 'BEGIN{total=0}{total+=$3}END{print total/NR}' $SOURCE_FILE` local memusedmax=`awk '{print $4}' $SOURCE_FILE|sort -n -r|sed -n 1p` local memusedavg=`awk 'BEGIN{total=0}{total+=$4}END{print total/NR}' $SOURCE_FILE` local memusedmin=`awk '{print $4}' $SOURCE_FILE|sort -n|sed -n 1p` local memfreemax=`awk '{print $2}' $SOURCE_FILE|sort -n -r|sed -n 1p` local memfreeavg=`awk 'BEGIN{total=0}{total+=$2}END{print total/NR}' $SOURCE_FILE` local memfreemin=`awk '{print $2}' $SOURCE_FILE|sort -n|sed -n 1p` local memcachemax=`awk '{print $7}' $SOURCE_FILE|sort -n -r|sed -n 1p` local memcacheavg=`awk 'BEGIN{total=0}{total+=$7}END{print total/NR}' $SOURCE_FILE` local memcachemin=`awk '{print $7}' $SOURCE_FILE|sort -n|sed -n 1p` local memused_cachemax=`awk '{print $4-$7}' $SOURCE_FILE|sort -n -r|sed -n 1p` local memused_cacheavg=`awk 'BEGIN{total=0}{total+=($4-$7)}END{print total/NR}' $SOURCE_FILE` local memused_cachemin=`awk '{print $4-$7}' $SOURCE_FILE|sort -n|sed -n 1p` local used_percent=`awk 'BEGIN{total=0}{total+=$5}END{print total/NR}' $SOURCE_FILE` echo "`date '+%F %H:%M:%S'`: Memory usage analysis" |tee -a $REPORTFILE echo -e "\t\t1.total memory: $memtotal MB"|tee -a $REPORTFILE echo -e "\t\t2.memory used: max=$memusedmax MB ,avg=$memusedavg MB ,min=$memusedmin MB"|tee -a $REPORTFILE echo -e "\t\t3.memory free: max=$memfreemax MB ,avg=$memfreeavg MB ,min=$memfreemin MB"|tee -a $REPORTFILE echo -e "\t\t4.memory cache: max=$memcachemax MB ,avg=$memcacheavg MB ,min=$memcachemin MB"|tee -a $REPORTFILE echo -e "\t\t4.memory used-cache: max=$memused_cachemax MB ,avg=$memused_cacheavg MB ,min=$memused_cachemin MB"|tee -a $REPORTFILE/usr/local/bin/gnuplot --persist < $GNUPLOTFOLDER/disk.$$ local SOURCE_FILE=$GNUPLOTFOLDER/disk.$$ plot_readiops="" plot_writeiops="" local nu=1 echo "`date '+%F %H:%M:%S'`: Disk Performance analysis" |tee -a $REPORTFILE for diskpart in `cat $file |sed '1,2d'|grep -v -i "device"|awk -F " " '{print $1}'|sort|uniq -d|sed '/^$/d'|grep -E "^[a-z][a-z][a-z]$"` do awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE >$GNUPLOTFOLDER/gnu_tmpfile.$diskpart plot_readiops="$plot_readiops \"$GNUPLOTFOLDER/gnu_tmpfile.$diskpart\" using 4 w l title \"$diskpart read IOPS\"," awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE|awk 'BEGIN{total=0}{total+=$4}END{print total/NR}'>$GNUPLOTFOLDER/t.$$ local read_avg_iops=`cat $GNUPLOTFOLDER/t.$$` rm -f $GNUPLOTFOLDER/t.$$ awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE|awk '{print $4}'|sort -n|sed -n 1p>$GNUPLOTFOLDER/t.$$ local read_min_iops=`cat $GNUPLOTFOLDER/t.$$` rm -f $GNUPLOTFOLDER/t.$$ awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE|awk '{print $4}'|sort -n -r |sed -n 1p>$GNUPLOTFOLDER/t.$$ local read_max_iops=`cat $GNUPLOTFOLDER/t.$$` rm -f $GNUPLOTFOLDER/t.$$ plot_writeiops="$plot_writeiops \"$GNUPLOTFOLDER/gnu_tmpfile.$diskpart\" using 5 w l title \"$diskpart write IOPS\"," awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE|awk 'BEGIN{total=0}{total+=$5}END{print total/NR}'>$GNUPLOTFOLDER/t.$$ local write_avg_iops=`cat $GNUPLOTFOLDER/t.$$` rm -f $GNUPLOTFOLDER/t.$$ awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE|awk '{print $5}'|sort -n|sed -n 1p>$GNUPLOTFOLDER/t.$$ local write_min_iops=`cat $GNUPLOTFOLDER/t.$$` rm -f $GNUPLOTFOLDER/t.$$ awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE|awk '{print $5}'|sort -n -r |sed -n 1p>$GNUPLOTFOLDER/t.$$ local write_max_iops=`cat $GNUPLOTFOLDER/t.$$` rm -f $GNUPLOTFOLDER/t.$$ awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE|awk 'BEGIN{total=0}{total+=$6}END{print total/NR}'>$GNUPLOTFOLDER/t.$$ local avg_read_throughput=`cat $GNUPLOTFOLDER/t.$$` rm -f $GNUPLOTFOLDER/t.$$ awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE|awk '{print $6}'|sort -n|sed -n 1p>$GNUPLOTFOLDER/t.$$ local min_read_throughput=`cat $GNUPLOTFOLDER/t.$$` rm -f $GNUPLOTFOLDER/t.$$ awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE|awk '{print $6}'|sort -n -r |sed -n 1p>$GNUPLOTFOLDER/t.$$ local max_read_throughput=`cat $GNUPLOTFOLDER/t.$$` rm -f $GNUPLOTFOLDER/t.$$ awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE|awk 'BEGIN{total=0}{total+=$7}END{print total/NR}'>$GNUPLOTFOLDER/t.$$ local avg_write_throughput=`cat $GNUPLOTFOLDER/t.$$` rm -f $GNUPLOTFOLDER/t.$$ awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE|awk '{print $7}'|sort -n|sed -n 1p>$GNUPLOTFOLDER/t.$$ local min_write_throughput=`cat $GNUPLOTFOLDER/t.$$` rm -f $GNUPLOTFOLDER/t.$$ awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE|awk '{print $7}'|sort -n -r |sed -n 1p>$GNUPLOTFOLDER/t.$$ local max_write_throughput=`cat $GNUPLOTFOLDER/t.$$` rm -f $GNUPLOTFOLDER/t.$$ awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE|awk 'BEGIN{total=0}{total+=$10}END{print total/NR}'>$GNUPLOTFOLDER/t.$$ local avg_await=`cat $GNUPLOTFOLDER/t.$$` rm -f $GNUPLOTFOLDER/t.$$ awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE|awk '{print $10}'|sort -n|sed -n 1p>$GNUPLOTFOLDER/t.$$ local min_await=`cat $GNUPLOTFOLDER/t.$$` rm -f $GNUPLOTFOLDER/t.$$ awk "(\$1==\"$diskpart\"){print}" $SOURCE_FILE|awk '{print $10}'|sort -n -r |sed -n 1p>$GNUPLOTFOLDER/t.$$ local max_await=`cat $GNUPLOTFOLDER/t.$$` rm -f $GNUPLOTFOLDER/t.$$ echo -e "\t\t$nu.$diskpart performance:"|tee -a $REPORTFILE echo -e "\t\t\t read iops:\t\t max=$read_max_iops ,\t avg=$read_avg_iops ,\t min=$read_min_iops"|tee -a $REPORTFILE echo -e "\t\t\t write iops:\t\t max=$write_max_iops ,\t avg=$write_avg_iops ,\t min=$write_min_iops"|tee -a $REPORTFILE echo -e "\t\t\t read data per second:\t max=$max_read_throughput KB,\t avg=$avg_read_throughput KB,\t min=$min_read_throughput KB"|tee -a $REPORTFILE echo -e "\t\t\t write data per second:\t max=$max_write_throughput KB,\t avg=$avg_write_throughput KB,\t min=$min_write_throughput KB"|tee -a $REPORTFILE echo -e "\t\t\t each io wait time:\t max=$max_await ms ,\t avg=$avg_await ms ,\t min=$min_await ms"|tee -a $REPORTFILE done /usr/local/bin/gnuplot --persist < $GNUPLOTFOLDER/network_sourcefile.txt local sourcefile=$GNUPLOTFOLDER/network_sourcefile.txt local titlerxpackage=`cat $file |grep -i "IFACE"|awk '{print $3}'|uniq -d` local titletxpackage=`cat $file |grep -i "IFACE"|awk '{print $4}'|uniq -d` local titlerxbyte=`cat $file |grep -i "IFACE"|awk '{print $5}'|uniq -d` local titletxbyte=`cat $file |grep -i "IFACE"|awk '{print $6}'|uniq -d` if [ $titlerxbyte == 'rxkB/s' ];then unit="KB" elif [ $titlerxbyte == 'rxbyt/s' ];then unit="byte" fi local rxpackage='' local txpackage='' local rxbyte='' local txbyte='' local nu=1 echo "`date '+%F %H:%M:%S'`: Network Performance analysis" |tee -a $REPORTFILE for netcard in `cat $file |grep -E "eth|em"|grep -v -i "average"|awk '{print $2}'|sort|uniq -d` do cat $sourcefile|grep $netcard>$GNUPLOTFOLDER/gnu_network.$netcard rxpackage="$rxpackage \"$GNUPLOTFOLDER/gnu_network.$netcard\" using 1:3 w l title \"$netcard $titlerxpackage\"," txpackage="$txpackage \"$GNUPLOTFOLDER/gnu_network.$netcard\" using 1:4 w l title \"$netcard $titletxpackage\"," rxbyte="$rxbyte \"$GNUPLOTFOLDER/gnu_network.$netcard\" using 1:5 w l title \"$netcard $titlerxbyte\"," txbyte="$txbyte \"$GNUPLOTFOLDER/gnu_network.$netcard\" using 1:6 w l title \"$netcard $titletxbyte\"," max_rxpck=` cat $GNUPLOTFOLDER/gnu_network.$netcard|sort -n -r -k 3|sed -n 1p|awk '{print $4}'` avg_rxpck=`awk 'BEGIN{total=0}{total+=$3}END{print total/NR}' $GNUPLOTFOLDER/gnu_network.$netcard` min_rxpck=` cat $GNUPLOTFOLDER/gnu_network.$netcard|sort -n -k 3|sed -n 1p|awk '{print $4}'` max_txpck=` cat $GNUPLOTFOLDER/gnu_network.$netcard|sort -n -r -k 4|sed -n 1p|awk '{print $5}'` avg_txpck=`awk 'BEGIN{total=0}{total+=$4}END{print total/NR}' $GNUPLOTFOLDER/gnu_network.$netcard` min_txpck=` cat $GNUPLOTFOLDER/gnu_network.$netcard|sort -n -k 4|sed -n 1p|awk '{print $5}'` max_rxbyt=` cat $GNUPLOTFOLDER/gnu_network.$netcard|sort -n -r -k 5|sed -n 1p|awk '{print $6}'` avg_rxbyt=`awk 'BEGIN{total=0}{total+=$5}END{print total/NR}' $GNUPLOTFOLDER/gnu_network.$netcard` min_rxbyt=` cat $GNUPLOTFOLDER/gnu_network.$netcard|sort -n -k 5|sed -n 1p|awk '{print $6}'` max_txbyt=` cat $GNUPLOTFOLDER/gnu_network.$netcard|sort -n -r -k 6|sed -n 1p|awk '{print $7}'` avg_txbyt=`awk 'BEGIN{total=0}{total+=$6}END{print total/NR}' $GNUPLOTFOLDER/gnu_network.$netcard` min_txbyt=` cat $GNUPLOTFOLDER/gnu_network.$netcard|sort -n -k 6|sed -n 1p|awk '{print $7}'` echo -e "\t\t$nu.$netcard load:"|tee -a $REPORTFILE echo -e "\t\t\t rxpck/s:\t\t max=$max_rxpck ,\t avg=$avg_rxpck ,\t min=$min_rxpck"|tee -a $REPORTFILE echo -e "\t\t\t txpck/s:\t\t max=$max_txpck ,\t avg=$avg_txpck ,\t min=$min_txpck"|tee -a $REPORTFILE echo -e "\t\t\t rxbyt/s:\t max=$max_rxbyt $unit,\t avg=$avg_rxbyt $unit,\t min=$min_rxbyt $unit"|tee -a $REPORTFILE echo -e "\t\t\t txbyt/s:\t max=$max_txbyt $unit,\t avg=$avg_txbyt $unit,\t min=$min_txbyt $unit"|tee -a $REPORTFILE nu=`echo $nu+1|bc` done/usr/local/bin/gnuplot --persist < $REPORTFILESysInfo $SYSINFO_FILEecho "" >> $REPORTFILECpuAllUsage $CPU_USAGE_FILECpuEachCoreUsage $CPU_USAGE_FILEecho "" >> $REPORTFILEMemoryUsage $MEMORY_USAGE_FILEecho "" >> $REPORTFILEDiskUsage $DISK_USAGE_FILEecho "" >> $REPORTFILENetworkPerformance $NETWORK_USAGE_FILEecho "" >> $REPORTFILE#脚本执行完成之后，会在性能数据所在的目录中生成#report.txt 性能报告文本#TotalCpuUsage.png CPU利用率图表#CpuCoreIdle.png 每个CPU核心的Idle#MemoryUsage.png 内存使用率#DiskIOPSPerformance.png 磁盘IOPS性能#NetworkPackagePerformance.png 网卡发包率性能#NetworkThougtputPerformance.png 网卡吞吐性能#################################################