will give you true fault tolerance. one of my drives developed an error a few months and the mirror detected the error but kept running. i replaced the drive 3 days later.
at the time, i was using intel matrix raid feature and had a 50gb mirror volume (raid-1) and a 370gb striped volume (raid-0) on two 250gb drives. even though 1 of the drives started throwing errors and the mirror volume was degraded, the striped volume was unaffected. i just copied the mirror and striped partitions over to the replacement drives.
i highly recommend using intel's matrix raid in 'regular' PCs.
I have experienced, as have many others, judging from questions posed in numerous forums, a hard drive failure in a RAID 1 array, followed by inability of the system to see the array, even though one drive is still good. This seems to be independent of hardware and OS. However, I have worked with HP/Compaq DL-series servers (W2k, 2003 server OS), and they don't miss a beat if a drive fails. It seems to be just the consumer-level stuff that has problems. My question is, does the RAID 1 spec itself stipulate that the data on the array should remain accessible after a drive failure, as most people, myself included, expect, or does it merely require the mirrored data structure on the drives, with operational fault-tolerance as an optional value-add? My suspicion is that the consumer vendors are offering 'RAID 1-ish' capability, rather than true RAID 1, but does the precise wording of the spec allow them to dance around this point? Any insights would be appreciated. Thanks.