Which Hot Spare will be used for a failed drive? – part 2

A while ago I talked about Hot Spares and how they are picked when a rebuild is necessary. It was almost 2 years ago and you can read it here.

Since then the rebuild / equalize technology has changed! Well, not for existing systems, but the new VNX family aka VNX2 does things a bit differently.

In the old days when a drive failed, a suitable Hot spare would kick in and the unprotected LUNs (regarding the failed drive) would be rebuild onto the Hot Spare. After a while, when the rebuild was done and the failed drive was replaced by a replacement drive, the data on the Hot spare would need to be copied to that new drive. This was called equalizing.

In the VNX2 (with MCx) this last step doesn’t exist anymore. So that means the Hot spare that was used to contain the rebuilt data is not longer a Hot Spare! It has become a regular drive! And that replacement drive will now be a new Hot Spare. When configuring a new VNX2 you’d see rules about Hot Spares and you simply don’t even need to configure Hot Spares anymore. Just make sure you have some unconfigured drives and you’re good. Your VNX2 will make sure they’re used as Hot Spares from then on.

If I remember correctly the DMX4 had a similar feature back in 2008, but it now flowed to the midrange platform as well.


  1. This new method makes more sense to me as a non storage admin 🙂 Less work is better.

  2. Yeah, but somehow we want to know where our data exactly is. So if these Hot Spares start “wandering around”, after a few replaced disks, we don’t know where our data is anymore.

  3. Permanent sparing doesn’t make any sense to me, as it totally nullifies vertical raid group design.
    Recently we added a few new VNX arrays and bumped into this … ‘feature’. When asked, EMC suggested to do a manual “copytodisk” after each disk replacement to keep the RGs spread over all DAEs.
    Damn you, EMC, at least give storage admins the option to chose between permanent and dynamic sparing 🙂

    • Vertical design? You have RAID Groups spanning DAEs? I wouldn’t do that. But if you meant you simply want to keep (private) RAID Groups together, I agree. Having the option to either choose for permanent sparing or the old fashioned way where data is moved to the HS first and copied back afterwards, seems valid to me. Why don’t you file an RFE? (= Request For Enhancement)

Would you like to comment on this post?

This site uses Akismet to reduce spam. Learn how your comment data is processed.