Expanding storage question/SOBR

[sosborne]

I have 2 x Dell PE R540 Win 2016 Standard Servers (one off site) with 2 x RAID 6 sets (11 x 10TB). One set attached to an internal RAID controller and the other to an external controller. When the servers were implemented I remember investigating SOBR but given the varying purpose and size of the backup jobs it didn't seem to be necessary especially given varying controller performance. However the repository I use for the large archive jobs is going to need extending very soon. This storage is attached to the external RAID controller and an extra enclosure is easily daisy chained off the first enclosure but from what I hear it is not advisable to extend RAID 6 sets to a large number of discs due to potential rebuild times. Now I can create an additional RAID 6 set (I don't think I can sell RAID 10 at this stage) so perhaps now is a good time to use SOBR. But I have a few questions:

1. When a SOBR repository is created how are the previous backup chains that are stored on the existing repositories managed when they are combined? And also the jobs that created them too?
2. (Bare with me here this ones somewhat convoluted) The offsite server has a few "backup chain" files which have not been directly created by a Veeam backup job/backup copy job but are Robocopies of backup jobs run at the main site (it seemed the best way to achieve the desired retention). These off site copies are reflected as "Backup Disk (Imported)" on the Veeam Backup Server albeit on the offsite repository. Now obviously SOBR is only a logical entity of Veeam so any Robocopies need to reflect a physical location change to the new drive at script level. But any future offsite copies should continue to appear as new restore points under the relevant original job in "Backup Disk (Imported)" once a SOBR repository has been created on the offsite server shouldn't it?
3. I know very little about Windows Storage Spaces but will it meet my needs better and do Veeam and it work well together?

[HannesK]

Hello,
first to mention: in my answers, I expect that SOBR is not stretched over two locations (reduce complexity). One SOBR only contains disks of one server / multiple JBODs / RAID-sets

@1: in the first place the data stays where it is. If you add a repository to a SOBR, it asks whether it should change all existing jobs that point to the repository to the SOBR. Not a big deal in general

@2: as I do not recommend to stretch SOBR over multiple locations, everything would be the same like before. I can suggest to think about using a backup copy job instead of robocopy. That could make things easier.

@3: In general, Veeam does not care about whether it runs on RAID or storage spaces parity. I personally would prefer the most reliable, easiest & fastest (in that order) option.

Best regards,
Hannes

[sosborne]

Thanks for the reply.

The intention was to have a SOBR repository on each site no split across a WAN.

I have just been trying to understand how past backup jobs and chains will be handled. And the future restore points of the offsite backups of mine.

I did originally look at backup copies but given the complexity of our retention requirements, especially long term, and the need for greater control over scheduling I could only use them on short term jobs.

[sosborne]

Any thoughts on RAID 6 vs 10 on direct attached storage? Obviously RAID 10 is better but more costly. But is it the only way to go for Veeam storage going forward? Previously we went RAID 6 for cost/capacity reasons. But with getting more informative advice from resellers who seem to know their stuff the potential rebuild times on 10TB disks make me nervous. However, it can be hard to justify an upgrade thats more expensive than the original server plus storage. Plus with already being committed you still need to make the best out of the investment so far, so I am wondering if going half-way with perhaps RAID 10 on new disks on the internal controller for short term jobs, and a 2nd RAID 6 set configured as a SOBR for archival is the way forward (I dread the thought of juggling the storage)?

Any thoughts appreciated.

[HannesK]

you already mentioned every pro / con from a software perspective both is fine.

the longest rebuild I have seen at a customer was almost two weeks for 8TB disks in a 12+2 RAID6 (to be fair, there was backup / copy load on the system almost all day long).