Short Version

Can I hot-swap disks in a Dell Compellent StorageCenter SAN with larger ones (one at a time, allowing them to rebuild between each change) and end up with more total available storage?

Little more detail

We have an old Dell Compellent StorageCenter SAN with 12x ST3600057SS (600GB) hard drives. Total available storage space is ~6TB. I want to increase the total storage significantly, replacing each 600GB drive with at least a 2TB drive.

The Compellent SAN has 3x logical volumes, all on the "High Priority (Tier 1)" storage profile (RAID 10 Active, RAID 5 Replay). 11 Disks are managed and 1 is spare (single redundant).

I know that with RAID 1/5/6, you can replace disks one by one and, once all are the new size, you'll end up with more usable space. However, I'm not sure if the same is true for RAID 10 or if there's anything special I have to do on the SAN.

Sadly I inherited this system with little to no documentation, so I'm kinda flying blind. I don't have any manuals for the SAN nor any documentation from the previous sysadmin.

Other possibly relevant info:

We have two model CT-SC040 controllers running version 6.6.5. RAID 5 stripe width is 9 Wide (90% efficient). RAID 6 stripe width is 10 wide (80% efficient).

  • 1
    RAID 5 stripe width is 9 Wide (90% efficient). RAID 6 stripe width is 10 wide (80% efficient). I wouldn't replace your 600GB drives with 2TB or larger ones given that RAID configuration. Those are almost certainly 15K-rpm SAS drives, which have much higher IO throughput than what are likely 7,200 or 5,400 RPM SATA drives. Such SATA drives likely handle less than 25% of the IO operations/second that 15K-rpm SAS drives can handle. The much lower IO throughput of the much larger SATA drives would make RAID rebuild times dangerously long. – Andrew Henle Dec 3 '17 at 16:26
  • @AndrewHenle Yeah, the original drives are 15K SAS drives. So if I don't really care about the RAID rebuild or about day-to-day IO throughput (we don't access all that much data - we should have just used a NAS from the beginning but I wasn't part of that decision), then I should be able to do that replacement though, right? Based on your response, I'm not planning on it. Now I'm just curious. – dthor Dec 4 '17 at 16:41
  • 1
    It's not really a question of supporting user IO. It's a question of a mostly-idle array when a drive fails going into rebuild mode. If it takes a week to rebuld the array (not out of the question), then during that time all the data on your array will be lost for good if another drive fails. Then restoring from backup will take a long time, too. What are you using for backups? If you have good backups and can handle the array being gone for a week (worst case) while restoring the data, go ahead and try the 2TB+ drives. – Andrew Henle Dec 4 '17 at 17:21
  • Ah, very good point. I hadn't thought of the "2nd drive failing during raid rebuild" case. I was only thinking of 1 drive failing. We have daily incremental backups and weekly full snapshots to a large NAS, as well as the replay on the SAN. While we do have the backups, I would not say that we can handle the array being down for a week. Thanks for the help. – dthor Dec 4 '17 at 17:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.