VM backup of SQL AlwaysOn AG clusters - backup of duplicate data

[minkee]

The current problem I'm facing is we are doing VM Image-based, SQL application-aware backups of our 3 node SQL AlwaysOn Availability Group clusters.

These VMs are huge and so are the respective data volumes with 15TB of data per VM node.

This is clobbering the backup infrastructure in terms of capacity as you can imagine. The thing is that the DB and Log drives are exactly the same across all 3 nodes so would it be possible to exclude the data (DB and log) drives from the backups on 2/3 nodes - apart from the single secondary node which is used to perform the SQL application-aware backup?

As this would save a massive amount of backup space...

[Andreas Neufert]

The issu with this approach is that you will not have a real disaster recovery for the system and you relay on the integrity of one of the servers. I saw so many situations over time where replication stopped working and you had only backups of this corrupt system then.

I would go maybe a different path and would look into deduplication.

You can use Veeam deduplication with Per-Job backupchain and include all 3 servers to the job. Potentially go a bit lower on the block size to optimize dedup.
You could use some software dedup from Microsoft as well or go with one of our supported deduplication appliances.
This would save you the space while continue to provide real disaster recovery.