What is a rule of thumb for the max average load for a single UNIX-based server before scaling?

Question

What is a rule of thumb for the average load for a single UNIX-based server before optimizing performance or scaling? For example, my server is running as shown:

load average: 0.44, 0.44, 0.43

possible duplicate of [How to understand the memory usage and load average in linux server](http://serverfault.com/questions/67759/how-to-understand-the-memory-usage-and-load-average-in-linux-server) — Tom O'Connor, Aug 08 '11 at 16:43

score 1 · Answer 1 · edited Apr 13 '17 at 12:14

1

Load is the number of processes waiting to run on the CPU. If it's below the number of CPUs available on that system, that means that, on average, you don't have any processes waiting.

A more complete answer is here.

edited Apr 13 '17 at 12:14

Community

1

answered Aug 08 '11 at 16:21

wfaulk

6,828
7
45
75

score 1 · Answer 2 · answered Aug 08 '11 at 16:27

This question can't be answered on a general basis. One example: A load average of 0.4 means something different with a single core system than with a 16 core machine, and it can also mean that your CPU is idling while processes wait for I/O on some OSes (i.e. Linux).

Do you experience any performance problems? Do you expect any increase in load?

What is a rule of thumb for the max average load for a single UNIX-based server before scaling?

2 Answers2