Hi I have a Windows 2003 SP2 server running on an ESX 3.1 server. When I log into the server, either via the VM Console or terminal services, the server is unbearably slow :( I checked out Task Manager, nothing appears wrong there. I opened System Monitor, and ran the default three counters... i) Pages/ Sec was constantly low ii) Avg Disk Queue length would shoot up to about 30 whenever I tried to do something on the server iii) % Processor was low Based on this, I assume that the cause is something to do with the disk, perhaps another virtual server on the ESX host is causing problems. Would I be correct? Are there any other recommended counters I can run to prove the issue is with the disk? I guess if the Pages/Sec were high as well, this would cause Avg Disk Queue Length to be high since the disk was being thrashed?

I'd look at disk reads/sec, disk writes/sec, avg disk sec/transfer, and %disk time. In addition you could look at running process explorer, http://technet.microsoft.com/en-us/sysinternals/bb896653.aspx When this runs, look at the line for the system process that tracks DPCs. DPCs take time away from the processor, but are not recorded in the % processor use counter and can indicate a high level of hardware activity. -- Mike Burr

Thanks Mike. >I'd look at disk reads/sec, disk writes/sec, avg disk sec/transfer, and %disk time. In addition >you could >look at running process explorer,# Are there any specific values above which indicate a problem relating to these counters?

Specific values usually depend on the benchmark of the server and the OLA for the service (set by the customer and service provider). I've given values where I've seen recommendations. For these, higher can be better (indicates better throughput if the disk is a bottleneck), but lower is preferred disk reads/sec disk writes/sec For these, lower is better disk queue lengths (anything sustained above 0 is a problem) avg disk sec/transfer (anything above .1 is possibly a problem, means that the average read or write is taking 1/10 of a second or more) %disk time (I've seen recommendations that this should be below 80%) DPCs/Interrupts (I haven't seen a hard limit here, but I've never seen it sustained above 10% on my system) -- Mike Burr