Veeam10: Scale-Out Repository - retention Policy

[mix]

Dear Users,

I updated my Veeam versione to v10. I have a Scale-Out Repository created and I configured a Backup Job with the Scale-Out Repository as Backup Target.
With version 10 I can configure GFS settings also in the primary Backup Job if I enable Full (Active or Synthetic) Full Backups.
In the primary Backup Job I configured 7 restore points as retention policy + the GFS settings + Synthetic Fulls on Sundays (the other 6 weekdays = incremental).
The problem now is that my performance Tier has to provide much more space (not only 7 restore points including 2 full backups due to the default behavior for incremental backups).

What do you suggest to use for my specific case? I need 7 restore points on the performance tier, the rest should be moved to the capacity tier.

p.s.: I guess reverse incremental would be much slower...right?

Thanks & Regards,
Jürgen

[veremin]

I think you can simply set 7 days as operational restore window. As soon as weekly backup is created, backup server will start moving previous restore points to Capacity Tier, giving you more or less the scenario you're after. Thanks!

[oleg.feoktistov]

I guess reverse incremental would be much slower...right?

GFS on primary jobs won't work with reverse incremental backup chains as described here.
Thanks!

[mix]

I think you can simply set 7 days as operational restore window. As soon as weekly backup is created, backup server will start moving previous restore points to Capacity Tier, giving you more or less the scenario you're after. Thanks!

I have now set 7 restore points on my primary backup job and 7 days on the scale-out-repository. The behavior is now that when I look at the filesystem level I find 1full + 6 incrementals and another 1full +5 incrementals on the performance tier. I think with the next incremental it will delete the first backup chain. If so, I need the double space on the performance tier?

[Gostev]

Only if you use legacy file system for your backup repository. ReFS and XFS repositories will not require the double space.

[mix]

OK, at the moment I use a NFS Share where the storage itself exports the share. In order to use ReFS or XFS I have to present the LUN to VMware, create a Datastore, place a VMDK on that and use ReFS or XFS in the VM (OS). Isn't that a kind of single point of failure or bottleneck? I mean all my backup proxies point there (single location) to save the backups on the repo?

[Gostev]

Indeed, your suggested approach is not a good one. There are too many moving parts, and it complicates the recovery - because you'll need your VMware host to be online to be able to access backups!

So instead, most customers just mount the LUN to any Windows or Linux server via iSCSI initiator (and make it the backup repository server). This way, in case of a disaster, you can quickly mount this LUN to any computer that is still alive (including your notebook), and start restores immediately.