Disk Response Time
IOMeter is sending a stream of requests to read and write 512-byte data blocks with a request queue depth of 1 for 10 minutes. The disk subsystem processes over 60 thousand requests, so the resulting response time doesn’t depend on the amount of cache memory.
The first thing we can see here is that all the RAIDs are somewhat worse than the single HDD. This is logical because the controller cannot but introduce some latency. However, the RAID10 might have been better than the single HDD by choosing what disk in a mirror pair can provide the requested data faster. We have seen other controllers do that. The 8-disk arrays are all faster than the 4-disk ones which, in their turn, are better than the degraded arrays. The latter do well in this test. Recovering data from checksums has a notable effect only on the performance of the RAID6 with two failed HDDs: recovering that much data is a heavy load that provokes a reduction of the disk subsystem’s response.
We see an odd picture at writing. On one hand, the RAID0 and RAID10 arrays have predictable standings: the larger the total cache of the HDDs and controller, the lower the response time is. But on the other hand, there is no order in the results of the checksum-based arrays. For example, the 4-disk RAID5 is inferior to the 4-disk RAID6. And the latter is even ahead of the 8-disk RAID6.