By "resilience", I was actually talking about the data, not the drive itself. RAID (5) gives me the confidence that I can lose a drive and still not lose the data.
I'm still thinking about the exact mechanics of this, because currently, I've got the comfort factor of the critical stuff on each server being backed-up to the other server by a scheduled backup, and then the nearline and offline backups supporting that. But if I consolidate both old servers into one new one, I lose that. The solution may well be a NAS box to give the scheduled backups in place of one server.
As for how to maximise drive life, good point. A long time ago I was involved in some tests with a US East Coast bank. They did extensive testing and error rates showed that their PCs lasted a lot longer if left on24/7 (with monitors powered down) than they did if turned off at night and turned on the following morning. This was put down to the mechanical stresses of spinning up drives, and the voltage surge when powering up circuitry.
However, it was years ago and before drives started powering themselves down internally, and when MTBF was a lot lower than it is today, and when technology was much less polished (drives, for instance, were 20MB, and I DO mean MB, not GB). So I'm far from convinced that what was true then is necessaily still true today.
I take onboard what you say about Raptors. I'll give it some thought, but I suspect you're right and it's the way to go.
Thanks, Zak.