Moving from the concepts of traditional Raid to ZFS, I’m trying to get a clearer understanding of performance when designing an array. I am familiar with setting up a traditional Raid10 across say 12 drives, so 6 mirrored pairs that are then striped, which would theoretically yield 12x Read Speed, and 6x Write Speed. If I lay it out the same way in TrueNas, 1 mirrored vDev then repeat data VDev 6 times (in the GUI), do I end up with 6 sets of mirrors that are then striped, similar to the traditional Raid10 I described? Simulating it in the VM lab it would “seem” that way. The follow on question is does it yield the same theoretical 12x read / 6x write performance?
Here is the answer from the people over at IX Systems:
Awesome! Thank you for the link.