0

I am trying to find out ideal load average for one of my linux servers. I understood load should be proportional to the no of cpu cores. But my doubt is that, how will we analyze this in case of multi core CPUs?

[root@server unnik]#  getconf _NPROCESSORS_ONLN
16
[root@server unnik]#


[root@server unnik]# grep "cpu cores" /proc/cpuinfo
cpu cores       : 4
cpu cores       : 4
cpu cores       : 4
cpu cores       : 4
cpu cores       : 4
cpu cores       : 4
cpu cores       : 4
cpu cores       : 4
cpu cores       : 4
cpu cores       : 4
cpu cores       : 4
cpu cores       : 4
cpu cores       : 4
cpu cores       : 4
cpu cores       : 4
cpu cores       : 4
[root@server unnik]#

So, suppose If I see a load average of 20, Should I compare it with 16(no of CPUs) or 16*4=64 (no of cpu x no of cores per cpu) ?

Improve this question
4
  • Your OS doesn't (usually) distinguish between logical and physical cores. There is no ideal load average without further parameters. If you want to save energy it would be as close to zero as possible. If you wanted to utilize it as much as possible it would be > something. In addition you will have to pay close attention to where you're reading your average from as they might mean different things.
    – Seth
    Jul 28, 2017 at 11:51
  • that doesnt really answer the question.
    – Unnikrishnan
    Aug 2, 2017 at 5:43
  • Which is why it's a comment. As you have not provided any information which tool you're using to get the average it might be dependent on that. If you're really unsure how you should be interpreting it (and the documentation doesn't have anything on it) built a testcase (with a defined load) and have a look. It's not really sensible to base it on the number of CPUs as the cores are the parts that actually do the work. But depending on the tool you might get different results htop would give you a limited visualization which might help you.
    – Seth
    Aug 2, 2017 at 6:11
  • In addition, as detailed in the question How do I read htop, the load of a system is different from the utilization. Only the second requires the number of cores. Depending on the kind of processes you're executing a high(er) load is to expected. The ideal load would be as low as possible as it would indicate as few as possible processes are waiting.
    – Seth
    Aug 2, 2017 at 6:13

0

Reset to default

You must log in to answer this question.