RAID-5 performance drop due to unaligned partitions?

    • OMV 4.x

    This site uses cookies. By continuing to browse this site, you are agreeing to our Cookie Policy.

    • RAID-5 performance drop due to unaligned partitions?

      (Disclaimer: I'm not using mdraid's RAID 5 or 6 at all so nothing to test or verify -- just curious).

      Here a Helios4 user claims the way OMV creates an mdraid (not using partitions but the whole devices) will lead to lower throughput compared to manually creating aligned partitions first and then creating an mdraid out of the partitions:…findComment&comment=69153

      Any thoughts on this?
    • To the time the OMV code was written it was common practice to use the whole device. I'm open to change that if this has been changed and the use of aligned partitions is the common practice nowadays. But this shouldn't be done based on the experience of only one user. A guideline from the MD maintainers or something else would be helpful here.
      Absolutely no support through PM!

      I must not fear.
      Fear is the mind-killer.
      Fear is the little-death that brings total obliteration.
      I will face my fear.
      I will permit it to pass over me and through me.
      And when it has gone past I will turn the inner eye to see its path.
      Where the fear has gone there will be nothing.
      Only I will remain.

      Litany against fear by Bene Gesserit
    • votdev wrote:

      this shouldn't be done based on the experience of only one user

      Absolutely agree especially since the numbers look suspicious anyway. ~76 MB/s write performance with 1 MB block size on a RAID-5 made out of 4 Ironwolf is way too low. But ~104 MB/s with 'optimal alignment' seems still too low. The tested filesystem was btrfs so further tests with ext4 and XFS would be needed anyway to catch mismatches between fs block size and RAID chunk size and so on...

      Edit: here some RAID numbers collected by the vendor himself with partitions and Marvell's XOR engine active:…findComment&comment=39961 -- ~80 MB/s at 1 MB block size.

      The post was edited 1 time, last by tkaiser ().