From: Ed Schmollinger (no email)
Date: Fri Aug 01 2003 - 11:14:28 EDT
On Fri, Aug 01, 2003 at 03:04:34PM +0200, J.A.J. van Belkum wrote:
> I understand that LA is the following:
>
> The average number of processes that the scheduler has to choose from
> when it decides which process is going to run next.
In the vernacular, load average is the average number of processes in
the run queue, which is almost what you said. (LA includes jobs that
are blocked on IO, too.)
> So (simply put) if you have 1 processor with a LA of 2 (so you have an
> average of two processes waiting to be run), and you add a processor....
> One process will go to the new processor and your LA will drop by 1.
> Resulting in a LA of 1. So if you have a load of 746, by adding 745
> processors your load will go down to 1.
This calculation does not match what you said above, but it is the
calculation used by some operating systems (HPUX being one example.)
Others use the overall run queue length, without dividing by
number_of_processors (Solaris does it this way.)
-- Ed Schmollinger -
|
|
|