SQL Server Availability Group backup single copy of data

[ian6754]

Hi, we have an availability group that has three replicas, one primary and two secondary. If we add all three instances are into the same backup job and enable application aware processing will Veeam backup all the databases on all three instances or is it clever enough just to backup the data once as the data is the same on all three replicas and there isn't much point in backing up the same data 3 times? At the moment we are protecting the data using SQL Maintenance plans which are intelligent enough to backup the data once. As the databases are rather large I don't want to triple the amount of storage we need in the backup repository if we move to using the native Veeam capability.

[foggy]

Hi Ian, since each node is a separate VM and Veeam B&R is an image-based backup solution, all three nodes will be backed up in their entirety. However, if you put all the nodes in a single backup job (per the best practices, for transaction logs processing and restores to work properly) and provided per-VM backup chains are not enabled on the repository, similar data will be deduplicated within a backup file so it wouldn't require triple the amount of space.

[ian6754]

Thanks foggy, excellent information and I had not figured in dedup. Good call. I guess that makes me still concerned about the amount of unnecessary processing that has to happen on the front end to send all the data to the back end before it is deduped. Is there anything in the pipeline from Veeam to mirror the way the native backup works where there are options in AG to just backup one of the replicas and select whether that is the primary or one of the secondarys? That would be neat as an option. Can you also just confirm that CBT is active on the backups and so each backup cycle we are actually only backing up changes since the previous one, albeit multiple times, and then deduping those streams? If so, that makes it an interesting choice as if we go with native SQL backups to disk and then back those up we will be backing up the whole database each time so there are definite advantages if CBT is being used, although I guess the merge or the synthetic backup will then have to run for each replica and dedup all over again. I'd like to have the best of both worlds where we can select to just backup one of the replicas AND do it using CBT. That would be nice.

[foggy]

Is there anything in the pipeline from Veeam to mirror the way the native backup works where there are options in AG to just backup one of the replicas and select whether that is the primary or one of the secondary?

No, there are no such plans.

Can you also just confirm that CBT is active on the backups and so each backup cycle we are actually only backing up changes since the previous one, albeit multiple times, and then deduping those streams?

Yes, CBT is always in action, if enabled (which is by default).