> I don't see an affordable solution to this. One can always use a
> single controller, but with decent disks it'll be saturated. And
> having a single disk per controller will run out of slots.
For resilience to controller failure, theoretically:
2 disks, 2 controllers, RAID-1 (50% of disk capacity)
3 disks, 3 controllers, RAID-5 (67%)
4 disks, 2 controllers, RAID-0 + RAID-1 (50%)
6 disks, 2 controllers, RAID-0 + RAID-1 (50%)
6 disks, 3 controllers, RAID-0 + RAID-5 (67%)
9 disks, 3 controllers, RAID-0 + RAID-5 (67%)
9 disks, 3 controllers, RAID-5 + RAID-5 (44%, but can survive failure
of at least 3 disks)
9 disks, 4 controllers, RAID-0 + RAID-5 (75%)
9 disks, 4 controllers, RAID-5 + RAID-5 (50%, but can survive failure
of at least 3 disks)
12 disks, 3 controllers, RAID-0 + RAID-5 (67%)
12 disks, 3 controllers, RAID-5 + RAID-5 (50%, but can survive failure
of at least 3 disks)
12 disks, 4 controllers, RAID-0 + RAID-5 (75%)
12 disks, 4 controllers, RAID-5 + RAID-5 (50%, but can survive failure
of at least 3 disks)
16 disks, 4 controllers, RAID-0 + RAID-5 (75%)
16 disks, 4 controllers, RAID-5 + RAID-5 (56%, but can survive failure
of at least 3 disks)
and so on...
Of course, this assumes that the kernel doesn't get hung by a
controller failure.
Cheers,
Bruno Prior [EMAIL PROTECTED]
