Help me understand Synthetic Full Backups

[jaybios]

Hi,

I'm struggling to get my head around how my cloud backups operate in our backup setup in relation to synthetic full backups.

We're running Veeam with direct attached storage, ReFS data store, a tape loader, and backblaze B2 cloud storage using immutability. I've got a scale out backup repository setup, and i do weekly backups to tape. Backup jobs do multiple daily incrementals, an active full once a month and synthetic fulls every weekend in between.

Prior to setting this up i was asking the question (everywhere but here it seems) as to how the synthetic fulls would work with cloud storage and I was told a synthetic full would use up the same storage space on the cloud tier as an active full, however since enabling it that doesn't appear to be the case, it only uploads data the same size as an incremental, even though in all the restores scenarios veeam says it's a full.

How does this work in terms of backup file dependability? Is that synthetic full still a standalone backup i can use to restore everything if the prior active full was deleted? does it still depend on the original active full if restoring from the cloud? What happens to a synthetic full normally if the preceding active full is removed?...as when they go to tape, it uses the same storage space as an active full, which is as i expected in order for it to be a standalone full restore capable backup, so i think i was expecting this from the cloud side too.

Hope i've explained that clearly enough...

[Dima P.]

Hello jaybios,

With object storage added to Scale-Our Repository as Capacity tier Veeam B&R ensures that similar blocks are not uploaded twice, as a result you see smaller upload stats even after synthetic or active full backups. Say, if active full backup consists of blocks A and B and next synthetic full backup consists of blocks B and C. Only block C will be uploaded during the synthetic full backup transfer. When active full gets removed, Veeam B&R recognized that block B is required for Synthetic full backup and remove only block A.

How does this work in terms of backup file dependability?

In object storage backup files are organized differently: set of blocks from the backup file are stored in the structure of directories.

Is that synthetic full still a standalone backup i can use to restore everything if the prior active full was deleted?

Yes.

does it still depend on the original active full if restoring from the cloud?

No.

What happens to a synthetic full normally if the preceding active full is removed?

Nothing. It will be stored in the repository as usual.

...as when they go to tape, it uses the same storage space as an active full, which is as i expected in order for it to be a standalone full restore capable backup, so i think i was expecting this from the cloud side too.

Tape, unlike Capacity Tier, writes files as files and any backup will consume it's actual file size (minus hardware compression). Thanks!