VPC Performance
Isaiah Williams (and others) commenting on my last post asked about the role of Ghost in this type of testing beta software.
Ghost is a good fit in this environment but I only use it as a fail-safe recovery mechanism. I think using VPCs for testing is way more convenient than using a tool like Ghost, particulary when you can also enable undo disks and just instantly forget all the changes if you didn't like the outcome. I use a series of base vpc images for all testing.
RAM is super important but the other thing is hard disk speed. Most of the problems notebooks face using VPCs is they're using 4200 or 5400 rpm drives. 7200 rpm drives (or 10,000 rpm drives on desktops) make a huge difference in VPC. It tends to hit drives pretty hard.
Other good options are external firewire or usb2 drives, for those with notebooks stuck with slow internal drives. I've got a Lacie 250G 7200rpm with 8G cache. Performs very well as long as your usb2 drivers are fine.