We must first of its profits; always want to run a variety of applications, you can use server monitoring tool to see how their server status in the server above; if you need to learn how server performance, you need to measure pressure tool;
- View server CPU core usage:
top
- View server memory usage:
free -m
- View server disk IO usage:
iotop
- Check the server disk usage:
df -h
- View server network IO usage:
iftop
- Upgraded version of the top command:
htop
Pressure measurement
- Tools for different types of server resources will be more pressure measured; they will just write use;
- CPU Stress Test:
# 使用lookbusy ./lookbusy -h # 打印帮助信息 ./lookbusy --n=4 --cpu-util=80 --mem-util=4096MB # 将4个逻辑核的使用率变为80%,内存使用4096MB
- Memory Stress Test:
# 使用stress-ng # 下载与安装stress-ng wget http://kernel.ubuntu.com/~cking/tarballs/stress-ng/stress-ng-0.08.01.tar.gz tar -xvf stress-ng-0.08.01.tar.gz cd stress-ng-0.08.01 make && make install # help stress-ng --help #在4个CPU逻辑核上面产生80%的负载(不耗用内存利用率): stress-ng --cpu 4 --cpu-load 80 #在4个CPU逻辑核上面产生20%的负载,并使用两个进程总共占用1024MB内存: stress-ng --cpu 4 --cpu-load 20 --vm 2 --vm-bytes 1024M #此时可以查看系统的CPU使用率,发现约为70%左右,因为在分配并耗费内存资源的时候,两个进程对两个CPU逻辑核的利用率是100%,因此:综合利用率是100%*0.5+20%=70%
- Disk IO performance testing tools: fio, dd, iozone
- Network bandwidth testing tools: Use netperf communications network between two machine performance testing
- netperf comprising two components:
- Client netperf
- netserver
#物理机上测试,在服务器端和客户端安装程序 wget https://github.com/HewlettPackard/netperf/archive/netperf-2.7.0.tar.gz tar -xvf netperf-2.7.0.tar.gz cd netperf-netperf-2.7.0/ ./configure && make && make install # 安装完成 # 在服务器端运行: netserver # 在客户端运行 netperf -H $ip -l 60 -t TCP_STREAM # ip是服务器端的ip地址
- netperf comprising two components:
reference
stress-ng
grasp of Linux PC performance benchmark tests
using NETPERF test network performance
network testing tool installation Netperf
Original: Big Box server Pressure Measurement and Monitoring Tools