> How could you test performance with and without a pagefile?
Create a system managed page file. Use the computer for a while doing the
things you normally do. Disable the page file. Use the system for a while
doing the things you normally do. Every time I've tried it, it was very
obvious that a page file was needed. I've tried it on many systems, 32 bit
and 64 bit, clients and servers, many versions of Windows, with varying
amounts of RAM up to 16 GB. I've never found a situation where disabling the
page file didn't noticeably cause problems and eventually slow the system
down. With a lot of RAM on a server dedicated to one task it was not
noticeable until the server had been up for several days. With win2k, XP,
and Vista it's usually noticeable almost immediately.
MS-MVP - Windows Desktop Experience: Systems Administration