Understanding system load

To Nha Notes | March 10, 2021, 9:58 a.m.

Let's clarify what is system load system term in Linux. The system load is number of processes which is actively using CPU at a given time, the waiting processes is also counted. When a process started, It increased by 1, and decreased by 1 when process terminated. For example if there is 1 process is using CPU, and two others are waiting, the system load count up to 3.

Because the system load can quickly increase or decrease based on execution time of processes, so we should use calculate average value for a period. So that we have the metric called Load average. We can see this value in Linux via commands like uptime, top.

Ex: load average: 1.44, 1.28, 1.27

• The first value 1.44 is load average for last 1 minute.
• The second value 1.28 is the load average for last 5 minutes.
• The third value 1.27 is the load average for last 15 minutes.

So, If there is 1 active process is running in single logical CPU server, it means server is 100% busy. It will be 50% busy in the case server have two logical CPUs core.

In best practices, this value should be <= 1 per a logical CPU core. So, depend on how many logical CPUs, we can specific which is suitable value for it.

Reference more in this blog: https://blog.appsignal.com/2018/03/28/understanding-system-load-and-load-averages.html