There is a *lot* of mis-information about raids out there!
Myth: After replacing a failed disk, the RAID array, will automatically re-build
Nothing could be further from the truth! A failed array must be manually re-built by the system administrator - the process once begun is largely automatic. Why the human intervention? Lets say someone wants to steal my data - if the array automatically rebuilt itself, all they would need to do is walk in with 3 drives and some time . . . There are RAID solutions involving a 4th "hot spare" - typically higher end SCSI controllers with battery backed up cache . .
Did you realize that ATA or SATA drives actually have the "controller" built on the drive? The ATA and SATA cards or built in chipset on the mainboard are nothing more than BUS controllers! This make RAID effectively software driven as all "intelligence" is actually performed at the O/S device driver level. Contrary to opinion Windows RAID is actually pretty good - an advantage of it is that it is instrumented, meaning that any problems or disk failures will be reflected in the Windows system event log. If you are considering an ATA or SATA solution, make sure it supports WMI, the better Adaptecs do.
A USB RAID is a *bad* idea, forget it
Consider an external NAS solution - Advantages are complete independance from workstation O/S, ability to share access to multiple machines. Disadvantages are access speed - currently limited to GB-Ethernet througput, a number of users are using several vendor's NAS solutions - I'm currently building my own based on FreeNAS ( http://freenas.org
Don't consider RAID an alternative to backups
Finally, if you choose a RAID solution, be *sure* you know how to recover a failed disk, and monitor ongoing health, I've seen too many users initialize their array after replacing a drive
hope this helps -John