I was looking at some linux server the other day, and was taking a look of the load averages using the “top” command. After staring at the load averages for a while, I thought to myself ‘What does that mean exactly?” and “How do they get those numbers?”. Well after some research I have found out that the load average is not a % of cpu usage, but is a representation of the load on the CPU by processes waiting for their time to use the CPU.
Below is a screenshot of the top command for reference. Notice in the top right are the load averages “1.23″ 1.17″ “1.05″. The first load average is for 1 minute, the middle is for 5 minutes, and the last one is for 15 minutes.
Now the important thing to remember about Unix/Linux load averages is that it is not a % of cpu usage, but how many processes are waiting to use the cpu. The higher the load average, the more load the server in under. Right about now you are probably wondering ” well what is a normal load average?” or “What is a high load average?”, the easiest way to determine if your load average is normal or high is to look at it in the following way:
| CPU Cores | Normal Load Average |
| 1 | 1.00 |
| 2 | 2.00 |
| 3 | 3.00 |
| 4 | 4.00 |
| 8 | 8.00 |
| 16 | 16.00 |
As you can see a normal load average for a single core cpu is 1.00 while a dual core cpu can handle a load average of 2.00. If it helps you out, you can think of 1.00 to be %100 for a single core and 2.00 be %100 for a dual core, and so on.
Above is a screenshot of top showing each individual cores ultilization. This is an 8 core cpu, so it’s normal load average is 8.00. You can see this output in top by pressing “1″ on the keyboard. Also, if you were wondering how I got top to display different colors, just press “z” on the keyboard.
The key thing to remember is that load averages in Unix/Linux are not % for cpu utilization, it is merely a representation of how many processes are waiting for their turn to use the processor.
References
http://www.linuxjournal.com/article/9001
http://blog.raamdev.com/2008/11/22/understanding-the-linux-load-averages


