Hi @dario72 -

As with about any question asked about one’s DR environment, the answer is “it depends” 
If you implement Hardened Repositories, you have no choice but to configure either a synthetic or active full in your jobs. So that’s an easy answer for that kind of job config.

Before I implemented VHRs, I used Forever Fwd method, and it worked just fine. If you implement Fwd with Synthetic Fulls, 2 things to keep in mind → you will need more storage because you have additional Full files; your ‘active backup chain’ will be less, which can be a good thing; and if you do implement Synthetic, it’s best to do so on ReFS (for Windows Repositories), or XFS (for Windows Repositories) so your storage space for the Synthetic Fulls is less (i.e. you don’t use up much more storage).

Hope that helps!

Additionally, if you do go with Forever Fwd, you run a transform operation at the end of each job run to merge data & “move” the original Full forward one spot in the chain to meet your configured Retention Policy, so the backup job itself will take slightly longer to complete. Just something to keep in mind.

@coolsport00 explained it really well. 

If you do decide to go with forever incrementals, enable Health Checks on the job to check for any corruption that might creep in.

https://helpcenter.veeam.com/docs/backup/vsphere/backup_health_check.html?ver=120

Ah yes..forgot about that @dips ...good add!  If FFwd is selected, aside from health check @dario72 , it’s good to select the defrag option as well, to keep your chain ‘clean’. 

Being SMB and not direct attached storage it would not be advisable to implement Synthetic Full backups as it would take longer depending on your NAS model than on the mentioned XFS/ReFS file systems that use block cloning.  If your NAS is powerful enough you could implement it but to me this would be a “try it and see how it goes” type of thing.  I would create a new backup job as a test and then implement the Synthetic Full to run after couple backups to see the performance and if not acceptable then you just delete the test job and data from the NAS.

Follow the other advice also given by Dips and Shane.

I agree with all the above postss.  On a NAS those synthetic operations might take a while depending. 

There is no issue with either backup method. I currently run FFWD backups on some very large (40TB+) servers and have never had an issue.

Make sure to enable health checks or sure backup to test every so often with any backup method. 

I was just going to recommend Active Full’s rather than synthetic because synthetic operations will not be efficient due to the data having to be read from the NAS and then rewritten to the NAS.  An active full would only send the data to the NAS, although from the production source, so note that there could be an impact there.  Forever forward is fine, but you do need to make sure you have maintenance enabled to prevent any fragmentation of the backup files as well as reclaim space and such.  If you’re using large VM’s, this process can take a long time and prevents other backups from being run during the time that maintenance is occurring, so I’d keep that in mind.

Good point @dloseke . I have defrag enabled on all of mine monthly otherwise performance will be poor. 

That is definitely the other downside using a NAS as SMB and doing synthetic fulls.  Great call out there Derek.