we r doing performance testing on a client server application – with 1 GB nt connectivity.
We have executed couple of our "low intensity" test cases, and our initial observations show that the CPU utilization on the application server is more than 90%.
We are using 4 CPU's and we do not employ load balancing. We are looking into individual CPU utilization now, like if not using load balancing is making any difference.
We think that CPU is our bottle neck, I am not sure what should be our immediate next steps......to make sure that CPU is in deed our bottle neck and if so what do we recommend to the management?
Do u folks have any suggestions? as what next steps needs to be followed? and what are the things that we need to look at?
Originally posted by MI_boy: We think that CPU is our bottle neck
<font size="2" face="Verdana, Arial, Helvetica">Thinking is not enough. You have to determine what is happening to make this conclusion. You are telling us that only now you start to look at the individual processors. This indicates to me that you have just done the very lightest of of investigations.
I suggest you first find out which process is causing this high usage. From there you can determin if the high processor usage is justified or not.
CPU utilization at 90% may not be bad. Are you seeing your run queues build up? If not, you may be alright.
Like Roland said, you need to know what is causing your bottleneck. If you think your CPU is your problem, you also need to look at your memory and your disks. Problems with these components can manifest themselves to look like CPU issues.
One thing you didn't mention was how close are you to meeting your performance requirements.