I’m torn between what I know (or think I do) and what Apple’s documentation states.
The hypothetical setup I’m considering is what is described as RAID 0+1 in Apple’s documentation for the Xserve RAID, which usually means a mirror of two striped sets comprising two identical drives. Like in this diagram: Minimal RAID 0+1 nested setup (Sorry, not allowed to post images yet)
Apple’s documentation along with the RAID Admin setup wizard, states that it can tolerate a single drive failure. Meaning, I assume, if two drives were to fail, there would be downtime, probably spent doing a full colonoscopy on the RAID controller, get new drives, get the backups out, etc, etc.
But isn’t this only in the worst case scenario that two corresponding drives from the two RAID 0 sets, e.g. Disk 0 and Disk 2 in the diagram, are fried? If Disk 0 and Disk 3 were to die simultaneously, shouldn't it be possible for the controller to rebuild the set? Hadn’t noticed this before.
Or does it mean that the diagram is an oversimplification? If you mirror a RAID 0 array with RAID 1, is there no guarantee that the individual drives would be mirrored?
I realize Apple’s enterprise days are long gone and people may not remember much, but if you have any insight on any other systems you’ve worked with, I’d love to hear your experiences.