Mixing active and synthetic full backups best practices

[filippoAdb]

In my backups job a synthetic full backup is created once a week, while an active full is created once a month.
I wonder if there are some recommendations about active full backups and their frequencies related to synthetic ones.
I suppose the main drawback of active full is that vm snapshots last long, so vm performance is somehow impaired.
This can be a no problem during the days (weekends in my case) when the workload is low. Are there other reasons to prefer synthetic over active full ?
What if I have a deduplicating repository for backups such as Hp storeOnce ? Does it change the validity of some considerations ?

My concern about using only synthetic full is that an error (a bad block of data) of the past last forever, while a full active would wipe it out.

[Mildur]

Hello Filippo

There is no general best practice.
For ReFS or XFS repositories (backed by a enterprise raid controller with battery cache), running synthetic full backups only is recommended. Synthetic full backups are space less on ReFS/XFS volumes. Active Full backups won't allow you to use space less full backups.
For NAS devices (SMB/NFS) used as backup repositories, we recommend Active Full only backups.

I suppose the main drawback of active full is that vm snapshots last long, so vm performance is somehow impaired.

Longer snapshot times, more pressure on your production environment, more backup storage required, more traffic goes through the network

What if I have a deduplicating repository for backups such as Hp storeOnce ? Does it change the validity of some considerations ?

For HPE StoreOnce, our recommendation is to use synthetic full backups.
https://helpcenter.veeam.com/docs/backu ... ll-backups

My concern about using only synthetic full is that an error (a bad block of data) of the past last forever, while a full active would wipe it out.

To protect you against such failures, it's important to use a stable and safe storage solution for your backup repositories. General purpose server with locally attached disks (enterprise grade raid controller with battery cache) or deduplication appliances are both trustworthy solutions for backup repositories.
To protect your recoverability against backup file corruption, make sure to follow our 3-2-1 rule.
- 3 Copies (1 production, 2 backups)
- 2 different media (use different repository technologies if you can)
- 1 backup copy offsite

Don't forget to have immutable or air-gapped copies of your backups.

Best,
Fabian