Ok so which blockhead at IBM decided that when presented with 10 disks and asked to make a RAID6 it would be a good idea to split it into two groups of five disks rather than a single 8D+2P RAID6? If you don't understand why a RAID5/6 array where the number of data disks is not a power of two is sub optimal go read up on it. TL;DR the parity calculations are much more efficient when you have a power of two. In my scenario this stupid decision by IBM has blown away 10TB of what should be usable storage.
Is there anyway to actually control the number of disks per RAID group given that unlike the IBM software engineers I actually have a clue about storage best practices?
Frankly given it is ZFS under the hood (something that IBM seem to want to keep quiet about), does anyone know why are we not creating RAIDZ arrays instead?
Is there anyway to actually control the number of disks per RAID group given that unlike the IBM software engineers I actually have a clue about storage best practices?
Frankly given it is ZFS under the hood (something that IBM seem to want to keep quiet about), does anyone know why are we not creating RAIDZ arrays instead?
