Hello,
Page faults can mean two things:
- the server is under stress
- or there is not enough RAM
Since 1996 I have installed many 100's of Windows Servers, but I never saw page faults connected to hardware issues.
Generally, having spikes of page-faults is normal - unless the faults continue at a high level. What is high, this depends a lot - and that is why Microsoft does not give us much guidance on that. On a single CPU single core P3 sustained over 20/sec is high (this was confirmed by Microsoft at the time). On a recent model dual-quad-core with SAS disks 100 or 200 sustained is not good. And I'm sure as soon as I hit submit, someone will disagree.
Just keep in mind that software like SQL gobbles up all memory at startup - so it can manage the memory itself, instead of letting the OS do the job. So from an OS point of view, 90% are always allocated- but does not mean anything; makes our troubleshooting more difficult.
As such, watch out for disk activity where the swapfile is, and as soon as the machine is swapping you will notice an increase in CPU as well.
The best way would be modulate the load on the server - so you can get some handle what "no load" and "lots of load" mean for your hardware and software configuration. Check MSDN under performance monitoring subject.
