[Techtalk] system utilization parameters

Julie txjulie at austin.rr.com
Wed Aug 6 07:56:27 EST 2003

Lena M wrote:
> Hello,
> thank you so much for answering my question yesterday.
> does any one of a good online resource/book about Redhat system utilization
> guidelines?
> For instance, what's an accepted cpu load average based on # of cpu,
> memory..
> what commands I can use to determine the above? I'm putting an advanced
> monitoring system for my client and would appreciate your input.

The short answer is "There isn't one."

Load average is related to quantity of work being done and inversely
related (generally speaking ...) to responsiveness.  So if your
machine is mostly doing non-interactive work a high load average
(within reason ...) means your machine is never wanting for real
work to do.  Unlike my machines which average 0.0 ;-)  On the other
hand, if your machine is mostly being used for interactive work, even
if "interactive" means "providing web services", a low load average
means it can immediately dedicate all available resources to whaever
task comes along.

As for what you can use, lots.  "uptime" will give you the average
number of jobs waiting on the run queue for CPU scheduling, "vmstat"
has memory usage info, "iostat" will give you disk I/O info.  You
can extract network information from "ifconfig", or frob /proc/net/dev
directly for the info.
Julianne Frances Haugh             Life is either a daring adventure
txjulie at austin.rr.com                  or nothing at all.
					    -- Helen Keller

More information about the Techtalk mailing list