On 5/3/06, Will H. Backman <[EMAIL PROTECTED]> wrote:
I'm looking for some hints on evaluating load average. I have a new
system that is showing load averages over .50 most of the time, but I
don't see that it is doing much according to systat vmstat. I figured
that this machine would be way overpowered for the job it is doing.
Is load average (like what is displayed in uptime) really a good
indicator?
What tips do people have for profiling?
-- Will
I've only really used load average in the context of what the load
average has been historically. It's not too great for tuning specific
performance bottlenecks (deciding it's time to get a faster disk, for
example) because too many components are involved in the calculation,
but if you happen to keep track of the typical load average
historically, and one day you notice that the load average is five
times higher than it ever has been in the past, you know something's
working harder than normal.
-Josh
