Raid 10 Would be better than 5, sort of, but as your drive count goes up you end up wasting drives.
4 Drives at 10GB each
Raid 5 = 30GB usable space. 1 10GB drive is your parity drive.
Raid 10 = 20 GB usable space. x2 drives are mirrors of the other two. Drives also have to be added to the array in even numbers.
6 Drives at 10GB each
Raid 5 = 50 GB w 1 10GB drive as parity.
Raid 10 = 30 GB usable space.
In either case, you have a 5th drive hanging out as a spare, either hot or cold.
I also think raid 10 has a bit of a performance hit over Raid 5. A write operation for Raid 5 is split across each of the drives.
So with say a file written out, imagine the file is QuikBrownFox
Disk No. and Data Fragment
5. checksum of the values in the blocks of the above. (remember, the bits for each disk block are numbers, the checksum allows for the system to calculate any of the drives.
With Raid 10.
Latency of the drives will be an issue as you can see each drive has more to write out. I'm not sure of the numerical calculations as a bottleneck on a given pro-sumer array.
But, the real bottleneck tends to be still down to the physical movement of the disks.
Oh, faster spinning drives will have less reliability for a given manufacturer. I also seem to recall seeing something that indicates the 3 and 6 gig drives having really awful reliability. I'm not sure why.
Edited by rmgill, 22 November 2018 - 1911 PM.