We have a 4 core RHEL virtual machine on VMware on which our Oracle database 11g is loaded. I wanted to understand if the load average shown by

uptime

command should be divided by 4 for a 4 core machine. In other words, can it be said that the load average can go up 4 to mean that there is 100 CPU utilization i.e. 4.0,4.0,4.0 for a 4 core be equivalent to 1.0,1.0,1.0 for a single core CPU?

I hope, I have been able to explain the problem.

Requesting a reply to my query.

Regards

My understanding has always been that you can go up to 4.0 for a 4 core machine.

My opinion also is that if you exceed 50-60% of your available processing limitations on a regular basis that you are in, or heading for, trouble. That being said, I'm not a server person, so not really used to high system strain on a regular basis.

1 members found this post helpful.

Not necessarily 100% utilisation as that will be a far more transient measure than the times used for load average.

It was explained to me as each core being a lane on a highway so a 4core is a 4 lane highway. In that case values less than 4 the highway is not busy, when it equals 4 exactly the highway is moving well but is at capacity and over 4 means that traffic is having to queue to get on to the highway.

If your load average is consistently at or nearly at the number of cores start looking at optimisation and maybe planning an upgrade soon. If it is consistently over then do the same but more urgently.

1 members found this post helpful.

fun fact: it can go way over the number of cpu's
http://stackoverflow.com/questions/2...e-vs-cpu-usage

cba to find the proper documentation on /proc/loadavg (the kernel source docs ? )
short story; it's the number of running processes
more precisely number of processes running times their running time divided by total time, in the last n minutes
so 100 processes always running (does not mean they are using the cpu) would give loadavg 100