Linux System Performance Evaluation

I. Check the number of CPUs, cores, and threads in Linux.

(1) Check the number of physical CPUs

Grep 'physical id'/proc/cpuinfo | sort-u | wc-l

(2) Check the number of cores

Grep 'core id'/proc/cpuinfo | sort-u | wc-l

(3) view the number of threads (logical cpu)

Grep 'processor '/proc/cpuinfo | sort-u | wc-l

The number of logical CPUs (possibly core, threads, or both) in each physical CPU:

Grep 'siblings '/proc/cpuinfo | sort-u

The command execution result shows that the server has 1 cpu, 6 cores, 2 cores, and 12 threads in total.

You can use dmidecode to view the cpu model. You can also find the specific cpu parameters based on the model.

Ii. Linux server performance evaluation (cpu, memory, disk IO)

(1) Factors Affecting Linux server performance

A. Operating System Level

CPU

Memory

Disk I/O bandwidth

Network I/O bandwidth

B. Application Level

(2) System Performance Evaluation Criteria

• Performance Impact Factors	Criteria
  	Good	Bad	Bad
  CPU	User % + sys % <70%	User %+ sys % = 85%	User % + sys %> = 90%
  Memory	Swap In (si) = 0 Swap Out (so) = 0	Per CPU with 10 page/s	More Swap In & Swap Out
  Disk	Iowait %< 20%	Iowait %= 35%	Iowait % >= 50%

  • Where:

    % User: the percentage of time when the CPU is in user mode.

    % Sys: the percentage of time when the CPU is in system mode.

    % Iowait: Percentage of CPU waiting for input/output completion time.

    Swap in: si, indicating page import of virtual memory, that is, switching from swap disk to RAM

    Swap out: so indicates page export of the virtual memory, that is, switching from RAM to swap disk.

    (3) System Performance Analysis Tools

    1. Common system commands

    Vmstat, sar, iostat, netstat, free, ps, top, etc.

    2. Common combination methods

    O use vmstat, sar, and iostat to check whether it is a CPU bottleneck.

    O use free and vmstat to detect memory bottlenecks

    O use iostat to check whether it is a disk I/O bottleneck

    O use netstat to check whether it is a network bandwidth bottleneck

    (4) Linux Performance Evaluation and Optimization

    1. Overall system performance evaluation (uptime command)

    [Root @ server ~] # Uptime

    16:38:00 up 118 days, 5 users, load average: 1.22, 1.02, 0.91

    Note: load average is the output value. The size of these three values cannot be greater than the number of system CPUs. For example, the system has eight CPUs in this output, if the three load average values are greater than 8 for a long time, the CPU is very busy and the load is high, which may affect the system performance, generally, the system performance is not affected. Conversely, if the output value of the load average is smaller than the number of CPUs, it indicates that the CPU still has idle time slice. For example, the output in this example, the CPU is very idle.

    2. CPU Performance Evaluation

    (1) Monitor the CPU of the system using the vmstat command

    This command displays brief information about the performance of various system resources. Here we mainly use it to view a CPU load.

    The output result of the vmstat command in a system is as follows: