All the RAID's I have set up at work (except one) have a parity drive and a global spare. If a drive gets dropped from the RAID, the global spare immediately starts the rebuild. In fact, this happened yesterday to my primary AD controller. The only downside to this is that after the failed drive is re-inserted or replaced, you still have to remember to assign it as the new global spare. The single server that does not have a hot spare is my backup server. Since I'm running CPS as well as Disk-Disk-Tape backups, it wouldn't be a big deal to lose the backup server. It can be rebuilt in hours from the disk image I pressed after install. The only thing that would be missing is all the CPS data. Again, not the end of the world.

All that being said, I HAVE had failures of both multiple drives at once, and twice of the controller card going belly-up. In all three cases, the data was completely gone, save for the backup.

I have also lost both regular SCSI controllers (Adaptec 1941) and IDE controllers (many types/brands). In both of scenarios, the data was still intact.

I have had MANY drives fail. I deal with it at least once a month. When you support as many PC's as I do, it's just a numbers game. It's GOING to happen. I'd say that drive failures vs controller failures is 100 to 1 or so, at least in my experience. Keep in mind, a controller has no moving parts. But what it DOES have, is RAM, which is where all my controller failures have occurred (except the IDE ones, which do not have RAM, per se).