Based on your description it seems that the performance you are getting from the RAID 10 volume is expected. This is due to the raid structure and the parity the volume needs to do with all the HDD in the array.
Hi there Kevin,
Unless my understanding is fundamentally flawed, there should be no parity calculation required for Raid 1+0 due to the fact it is simple mirroring and striping?
I have anyway though had one drive through a SMART 'drive is about to suffer meltdown' warning, which is odd as previous conveyance tests with smartctl did not throw errors, but still, I will replace the drive and retest.
My previous experience with RAID 1+0 did not show these massive dips in performance.
You are correct saying that the RAID 10 has no parity in the array. The performance you are getting on a raid 10 might be because the array is constantly checking and verifying the structure. Since this is a software raid controller, there is no way to change these settings.