- Advertisement -

In this article, we will explain one of the critical Linux system administration tasks – performance monitoring in regards to system/CPU load and load averages.

Before we move any further, let’s understand these two important phrases in all Unix-like systems:

+ System load/CPU Load – is a measurement of CPU over or under-utilization in a Linux system; the number of processes which are being executed by the CPU or in waiting state.
+ Load average – is the average system load calculated over a given period of time of 1, 5 and 15 minutes.

In Linux, the load-average is technically believed to be a running average of processes in it’s (kernel) execution queue tagged as running or uninterruptible.

Note that:

+ All if not most systems powered by Linux or other Unix-like systems will possibly show the load average values somewhere for a user.
+ A downright idle Linux system may have a load average of zero, excluding the idle process.
+ Nearly all Unix-like systems count only processes in the running or waiting states. But this is not the case with Linux, it includes processes in uninterruptible sleep states; those waiting for other system resources like disk I/O etc.

How to Monitor Linux System Load Average

There are numerous ways of monitoring system load average including uptime which shows how long the system has been running, number of users together with load averages:

- Advertisement -

The numbers are read from left to right, and the output above means that:

+ load average over the last 1 minute is 1.98
+ load average over the last 5 minutes is 2.15
+ load average over the last 15 minutes is 2.21
+ High load averages imply that a system is overloaded; many processes are waiting for CPU time.

We will uncover this in the next section in relation to number of CPU cores. Additionally, we can as well use other well known tools such as top and glances which display a real-time state of a running Linux system, plus many other tools:

Top Command

Glances Tool

The load averages shown by these tools is read /proc/loadavg file, which you can view using the cat command as below:

On desktop machines, there are graphical user interface tools that we can use to view system load averages.

Understanding System Average Load in Relation Number of CPUs

We can’t possibly explain system load or system performance without shedding light on the impact of the number of CPU cores on performance.

Multi-processor Vs Multi-core

+ Multi-processor – is where two or more physical CPU’s are integrated into a single computer system.
+ Multi-core processor – is a single physical CPU which has at least two or more separate cores (or what we can also refer to as processing units) that work in parallel. Meaning a dual-core has 2 two processing units, a quad-core has 4 processing units and so on.

Furthermore, there is also a processor technology which was first introduced by Intel to improve parallel computing, referred to as hyper threading.

Under hyper threading, a single physical CPU core appears as two logical CPUs core to an operating system (but in reality, there is one physical hardware component).

Note that a single CPU core can only carry out one task at a time, thus technologies such as multiple CPUs/processors, multi-core CPUs and hyper-threading were brought to life.

With more than one CPU, several programs can be executed simultaneously. Present-day Intel CPUs use a combination of both multiple cores and hyper-threading technology.

To find the number of processing units available on a system, we may use the nproc or lscpu commands as follows:

Another way to find the number of processing units using grep command as shown.

Now, to further understand system load, we will take a few assumptions. Let’s say we have load averages below:

On a single core system this would mean:

+ The CPU was fully (100%) utilized on average; 1 processes was running on the CPU (1.00) over the last 1 minute.
+ The CPU was idle by 60% on average; no processes were waiting for CPU time (0.40) over the last 5 minutes.
+ The CPU was overloaded by 235% on average; 2.35 processes were waiting for CPU time (3.35) over the last 15 minutes.

On a dual-core system this would mean:

+ The one CPU was 100% idle on average, one CPU was being used; no processes were waiting for CPU time(1.00) over the last 1 minute.
+ The CPUs were idle by 160% on average; no processes were waiting for CPU time. (0.40) over the last 5 minutes.
+ The CPUs were overloaded by 135% on average; 1.35 processes were waiting for CPU time. (3.35) over the last 15 minutes.

In conclusion, if you are a system administrator then high load averages are real to worry about. When they are high, above the number of CPU cores, it signifies high demand for the CPUs, and low load averages below the number of CPU cores tells us that CPUs are underutilized.


- Advertisement -

There are no comments yet

  • Hello, guest