Switching to backup copy to & replica from, SOBR

[pkelly_sts]

I currently run a backup copy to remote site (originally seeded) and then replicate from that backup copy to the remote sites VM infrastructure.

Having found the huge performance benefits of local backups to a Scale Out Backup Repository (SOBR) whilst also using per-vm backups, I would like to consider switching to this source backup as "primary" but will need to copy/seed it to the remote site.

I've seen topics in here discussing some subtle but important differences between seeding from/to simple & SOBR repositories so want to ensure I'm working from the correct information.

Lastly, I just found at https://helpcenter.veeam.com/backup/vsp ... ng_vm.html that replicas can not be seeded from SOBR. Is this likely to change in future? If I seed a new backup copy to the remote site, can I continue replication using that backup copy as a source or will I have to start replication again?

[foggy]

To seed from backup stored on SOBR, you need to copy contents of that particular job folder from all extents into a single folder with the same name on the target repository, rescan it and use for mapping. Provided everything is fine, you would be able to further replicate from this backup.

[pkelly_sts]

Ah, ok, so simplify it to a single folder first, can I then move it to a SOBR at the destination site after? Main reason I want to is to speed up transform/merge processes if possible.

[foggy]

In theory, yes. Though we haven't specifically tested this scenario, so I would appreciate if you report your findings if you try.

[pkelly_sts]

Hmm, one thought that has just occurred to me is that one of the benefits of the current backup is that I've split it across two repositories, but configured BOTH member repo's to hold both full and increments, for maximum performance. If I consolidate these to a single folder (so back up both folders, then restore both to a single folder at the destination), then I'll end up with everything sitting in a single repository again, thus losing the benefits of spreading across the two.

I'm guessing it's pretty hard to spread them out again once they're at their destination so how would I achieve it? Remember, this is a seeded backup copy to a remote destination so running an active full across the wire isn't an option.

(Bear in mind the whole reason for me switching to this split repo was to improve performance).

[foggy]

Again, in theory, it should spread the subsequent backups across both extents.

[pkelly_sts]

OK, sounds like I'll need to test the process with a smaller dataset just to confirm behaviour then - takes me a while to arrange seeds as we only visit the remote site once per week but I shall report back any findings here anyway...

[lowlander]

Can we also copy individual vbk and vibs to a single location for seeding ? Or should all content of all extents be placed in a new location ?


Verzonden vanaf mijn iPhone met Tapatalk

[lowlander]

Sorry for the confusion. I mean can we move a single vm backup


Verzonden vanaf mijn iPhone met Tapatalk

[foggy]

And one more time I need to say that in theory, this should be possible: get all the VM files from all extents into a single folder along with the VBM file, rescan that folder, open this backup properties and remove all missing backups, use the backup for mapping.

But again, we haven't tested this particular scenario, so you can try.