This 4-part series will explain six key performance metrics to monitor in a Linux server. By the end of this series, you will understand what factors in a Linux system you need to look out for that contribute to the overall system performance and also their relative importance.

Is your Linux server slow or are applications hosted on it showing unexpected behavior or signs of sluggish performance? Many factors in a Linux server affect its overall performance or the performance of applications hosted on it.

But there are a few key metrics you need to track that can impact the most critical things such as response time or execution time of processes, throughput – the total amount of work the server can do in a given time, and much more.

Let’s look at the first two key performance metrics in this first part of the series.

Table of Contents

Being the system’s brain, a properly functioning CPU is a critical part of your Linux server or any computer. Therefore CPU usage is one of the essential aspects to track in a Linux server, which is used to gauge system performance in terms of throughput.

There are several command-line-based and graphical user interface tools for monitoring CPU usage on a Linux system such as top, glances (my favorite), htop, and more.

Glances – Linux Monitoring Tool

Glances is an open-source real-time monitoring utility that monitors several aspects of your Linux system such as CPU, memory, disk, and network usage.

To install glances on your Linux distribution, run:

<code>$ sudo apt install glances             [On <strong>Debian, Ubuntu and Mint</strong>]
$ sudo yum install glances             [On <strong>RHEL/CentOS/Fedora</strong> and <strong>Rocky/AlmaLinux</strong>]
$ sudo emerge -a sys-process/glances   [On <strong>Gentoo Linux</strong>]
$ sudo apk add glances                 [On <strong>Alpine Linux</strong>]
$ sudo pacman -S glances               [On <strong>Arch Linux</strong>]
$ sudo zypper install glances          [On <strong>OpenSUSE</strong>]    </code>

The following is a screenshot of a section of the glances monitoring tool on a Debian Linux server showing CPU utilization statistics.

<code># glances</code>

Here the overall CPU usage is 29.5 percent, and more CPU time is utilized by user space processes or applications. It shows the total number of tasks (156), the number of tasks running (7), the number of tasks in sleep mode (81), and others. It also shows a list of processes by CPU consumption by default.

Top – Monitor Linux Processes

The top command provides a dynamic real-time view of all the running processes in the Linux system along with CPU and memory utilization.

<code># top</code>

htop – Linux Process Viewer

htop is an interactive Linux system process viewer and process manager that shows the information on all the running processes by the CPU and memory usage on a system.

To install htop on your Linux distribution, run:

<code>$ sudo apt install htop             [On <strong>Debian, Ubuntu and Mint</strong>]
$ sudo yum install htop             [On <strong>RHEL/CentOS/Fedora</strong> and <strong>Rocky/AlmaLinux</strong>]
$ sudo emerge -a sys-process/htop   [On <strong>Gentoo Linux</strong>]
$ sudo apk add htop                 [On <strong>Alpine Linux</strong>]
$ sudo pacman -S htop               [On <strong>Arch Linux</strong>]
$ sudo zypper install htop          [On <strong>OpenSUSE</strong>]    </code>

The following is a screenshot of a htop monitoring tool, which shows a list of running processes by CPU Usage.

<code># htop</code>

Based on the figures you have seen in the previous screenshots, the CPU usage metric is divided into three main categories or states:

• user – indicates the percentage of CPU time used by user space processes.
• system – shows the percentage of CPU time used by the kernel.
• idle – shows the percentage of CPU time not actively being used.

Additionally, there are also the following CPU sub-states:

• nice – is a subset of the user state which indicates the percentage of CPU time occupied by user-level processes with a positive nice value (lower scheduling priority than other processes).
• irq or hi – shows the percentage of CPU time spent servicing hardware interrupts.
• softirq or si – shows the percentage of CPU time spent servicing software interrupts.
• iowait or wa – a subset of the idle state that shows the percentage of CPU time spent waiting for I/O operations such as reading and writing to or from disk.
• steal – shows the percentage of CPU time taken by a virtual CPU in a hypervisor to involuntarily wait for a physical CPU for processing time.

Monitor Linux CPU Core Utilization

If your server has more than one processor (sometimes called a multi-core system), you can also monitor the utilization of each core. In glances, simply press 1 to have a view like the one shown in the next screenshot.

You can also run the following command to find the total number of processors in your server:

<code># cat /proc/cpuinfo | grep ^processor</code>

To find the top running processes by CPU usage, run the following ps command:

<code># ps -eo pid,ppid,cmd,%mem,%cpu --sort=-%cpu | head</code>

You might also like:

• Cpustat

The above is the detailed content of 6 Key Performance Metrics to Monitor in Linux Servers. For more information, please follow other related articles on the PHP Chinese website!