One of our physical w2k8 5.1 U1a vCenter servers is running with 96GB RAM and using around 40GB in normal operations (150hosts and 2500 VMs). Windows Page File Size is configured with Custom Values (2GB on C: and 16GB on another drive)
Almost on a daily basis, one of its services (SSO, SDK, etc) stops responding and only a restart of the service or the whole server resolves the issue. The server performance is slow anyway but we got used to it... 
Checking the server with process explorer shows an enourmus amount of Page Faults, generated by vpxd.exe (~220,000,000) and also java.exe (~5,000,000 it is the java.exe process with path to the vCenter directory) see attached file
Restarting the server takes 5 mins, but starting up the vpxd takes 20 mins! During this time, normal logging is going on with several warnings:
warning 'VpxProfiler' opID=SWI-266b9ebb] VpxUtil_InvokeWithOpId [TotalTime] took XXXX ms
Any ideas where to check further? This time, I am trying it with you guys before the support wants the usual Log Bundles and other (mostly useless) troubleshooting steps...