Solved

Benchmarking RAID1

Hi, I recently got a new system on which I set up RAID1. I got it for improved safety and, hopefully, read performance. I'm trying to find out how my RAID1 setup fares against using a single drive. Therefore, I want to run a benchmark with both drives, then disable one drive and run the benchmark again using a single drive. How can this be done? What I'm concerned about is that disabling one drive will inevitably cause at least some data to be written to the single drive, causing the drives to no longer be mirrored. How would you synchronize the drives again in this case, or a similar case where one of the drives fails?
8 answers Last reply Best Answer
More about benchmarking raid1
  1. Dont do that without a backup of your data.

    Why not just benchmark your RAID; you probably will only run sequential benchmarks; and these should be predictable. So you can see instantly whether your RAID driver accelerates RAID1 or that you're stuck with single-disk performance, as often the case with simple drivers.
  2. Sorry to say it, but the best time to do that sort of benchmarking is when you're setting the system up and before you've committed any critical data to it.

    You can remove one drive for testing, but it WILL no longer be a mirror. Whether or not you can put the drive back into the RAID set and rebuild it depends on the controller/software - I've seen posts that suggest some RAID subsystems will refuse to reuse a drive they consider to have been "failed".

    Again, the time to have experimented with pulling drives out, replacing and rebuilding them is when you first set the system up and are learning how the recovery procedures for your particular RAID controller work.
  3. I just turned on this PC for the first time about an hour ago, so short of Windows, there's absolutely nothing on here yet. While it'd be slightly annoying to reinstall, it certainly wouldn't be of any harm. So, if I understand you correctly, I can simply pull it out, plug it back in when it's no longer mirrored, and the RAID controller should work its magic automatically?

    sub mesa, I don't understand how I could benchmark RAID vs a single drive by only benchmarking RAID. Could you elaborate? That sounds like a more attractive way of doing this.
  4. Best answer
    Well then, you're in an excellent position to benchmark. Go ahead and do exactly what you described and see how it works. When you plug the drive back in the RAID controller *should* rebuild the mirrored set.

    The acid test would be to reinstall Windows with no RAID at all. That would give you a single-drive performance baseline that is in no way affected by the RAID software/hardware.

    This is an excellent opportunity for you to test your recovery procedures, and what your about to do is a big part of that. Make notes of the kinds of messages you get, exactly what you need to do to get everything working again, etc. They'll be invaluable when you have live, precious data on your RAID set and a drive fails.
  5. Reads? Something you can probably benchmark, but I wouldn't count on ever noticing a difference in everyday use.
    Just like right now, you are not noticing any difference at all, and are looking for a way to find out if there is. If there was a difference worthwhile, you wouldn't be asking the question!
    I don't care if it is RAID 1, back it up somewhere or you will be sorry one day. :)
  6. He's right though, if you rely solely on RAID1 to protect against data loss this is a poor solution. Its convenient though, since it requires little maintenance, but not the most secure.

    You may be better off just using the 2 drives as normal separate drives without RAID; and backup manually or using some task scheduler/backup program to do this for you. And you use the second harddrive as backup. This is much more secure, as it protects against more dangers and removes the raid layer which adds a point of failure in itself.
  7. Best answer selected by theBNich.
  8. jitpublisher, at the time, I didn't know if there may have been a difference or not--I hadn't used a single SATA3 HD yet before installing the RAID, so I had no basis for my comparison (other my previous drive which was far slower than the two drives I just bought). As it turned out, though, there was no difference; as expected, the onboard RAID controller just isn't good enough to make those kinds of optimizations.

    sub mesa, of course--I never intended to use RAID1 as my sole backup source; I just wanted it in case of an HD failure. I already have an external USB HD which I plan to use for backups; otherwise, your idea of using the second drive for backup only would have been a good approach.

    After buying two HDs and seeing no performance improvement in RAID1, I have instead converted to RAID0. I was hoping RAID1 would be a good compromise between safety and performance, but since it offered no performance improvements, I didn't feel like the added safety for the minimal risk of HD failure was worth it. With RAID0, I now have the full capacity that I purchased along with the speed boost when doing large file operations. Obviously, though, I'll have to make sure to take frequent backups (which I would be doing anyway) to avoid unforeseen data loss.
Ask a new question

Read More

NAS / RAID Benchmark Storage