load avg.

[Scott_20]

when you see load avg - top/uptime - is that for the entire system, just cpu's? need clarification

[Jakes Louw]

Top and uptime will report the latent demand for CPU resources.
But I have to question the way these figures are obtained when one considers we now have 64-way SMP servers, and I'm pretty sure the processing queue can be represented in a better way.

[Mic V.]

Load average represents the total number of runnable (i.e., not blocked or any other non-runnable state) processes in the system.

If you have a 6-CPU system, a rule of thumb is that you can generally expect to see a load average of 6 without experiencing "slowness."

However, it is not necessarily possible to correlate the load average with the responsiveness of the system; "it depends" on the nature of the processes that are running and system state/resource allocation. I've seen a single-CPU system with a load average in the 20s where the shell was still very responsive (rare case). I've also seen a 6-CPU system with a load average of 4 (implies 2 idle CPUs) where users were complaining aboout the length of time it took to run database jobs (the presumption being that there were database locking issues or application problems).

HTH,
Mic

[Graham Cameron_1]

From "man uptime"

... the average number of jobs in the run queue over the last 1, 5, and 15 minutes for the active processors.

-- Graham