in your experiences does of real world testing of RAID5 come "close" or "far" behind RAID0 in terms of performance as the underlying storage for a heavy write OLTP workload? Please for the sake of the question ignore any concerns of the usage RAID0 because it lacks durability, as this is out of the scope of the question.

It can be assumed the environment is virtualized in VMWare, the storage median is SAS SSD, the DBMS SQL Server 2017, the OS for the VM is Ubuntu, there is high bandwidth between storage and guest vm 10Gbps for network and 12Gbps for storage.

Thanks you

  • 3
    Your question is almost certainly going to end up being closed because it's much too broad. Try to chop it in smaller bits. RAID 0 in development is fine -- if you can tolerate the downtime when a disk fails. If not and it would result in many developers sitting around twiddling your thumbs, then your development environment is actually production for developers. Feb 5 at 19:06

> is this the best approach for performance?

That is at the same time too broad a question (i.e. will be very opinion based) and to narrow (other people looking for advice are unlikely to have that same config)

> Please remember downtime is acceptable for some servers at a time in the case of a drive failure because this is not production workloads

Are you absolutely 100% sure that this is acceptable? I would want it signed-off in triplicate from the very top because if one bad day you have two drive failures so two of those four RAID0 pods are out of action, you might have an angry manager breathing down your neck wanting to know why half the dev team are out of action (or at least experiencing delays) for how-ever long your time to restore is… To dev managers, dev workloads (and QA workloads) are production workloads. And to get it signed off you are going to need more information, for a start a pessimistic ("realistic worst case") idea of what the restore-to-100% time might be.

  • Thank you David for the reply. We do have sign off from the organizations top technical officer, so we are covered there. In this case I don't believe there to be a "right" answer, I am more seeking suggests on what could be done different with the same restrictions and the same goals in mind.
    – birchmeier
    Feb 5 at 19:02
  • Hey David, I updated the ask to include a request of others have done if they found themselves in similar situations. While I thank you for your response it is not helpful. While the concern of downtime should not be ignored completely I was more seeking responses to the effect of "you should try RAID6 with 8-24 drives because the performance is small and you'll get improved durability".
    – birchmeier
    Feb 7 at 20:45
  • @birchmeier - for better or worse, better IMO, StackExchange sites are intended for less open questions where there is the possibility of a definitive right answer, rather than requests for suggestions. Feb 9 at 10:40

Not the answer you're looking for? Browse other questions tagged or ask your own question.