v9 - Reverse incremental without active full

[EricB]

Hi guys,

I know that a lot have been said already about reverse incremental jobs and how they still require an active full job to maintain restore point integrity and keep the backup files in good shape.

Currently I'm running a V9 U1 deployment that makes use of NetApp SnapShots and backup from a NetApp SnapVault source.
The reason that I still require filling a Veeam repository is because unfortunately in the current Veeam release that is still a requirement in case you need to perform backups to tape as you cannot directly use the data in the NetApp snapvault.

Now the situation is as follows, if I would configure a reverse incremental job for filling the Veeam repo (which would be most ideal in this scenario as the repo is only a necessity and not a requirement for restore points) I would traditionally have to configure active full backups which would pull data again from the NetApp storage (i'm unsure if it would do this from the primary NetApp or if it would use the secondary SnapVault source again) but anyhow it would be slow as it only uses GbE connectivity (about 25TB of data to backup which results in about 10 TB for a single full after compression and dedup, so do the math... )

That got me thinking about disabling the active full entirely and use storage-level corruption guard to safeguard restore point integrity.
Veeam states the following about this in VMCE training: "Backup files produced by primary backup jobs can now be periodically scanned to identify storage issues, such as a bit rot. Corrupt data blocks are auto-healed by retrieving correct data from the production storage, increasing the reliability of forever-incremental backups and removing the need for periodic full backups."

I would like to use the Reverse always full restore point to use as a source for regular tape jobs without having the need for another synthetic full as that messes up my schedules.

Does anyone have any experience with this? Is it safe to use and can i really do entirely without full and or synthetic fulls?

[veremin]

You can disable full backups for reversed incremental jobs, as long as you enable health check and run periodic Surebackup jobs. Thanks.

[foggy]

Eric, am I right in thinking, that if not tape backups, you would be left with just NetApp snapshots and without any Veeam B&R backups at all?

[haslund]

Depending on your needs in regards to tape backup, just remember if you use Reverse Incremental then every backup to tape will be a full backup.
Having full backups sent to tape on every run, can cause high amount of tapes to be used.
If this is a problem for you, consider using forever forward incremental instead and virtual synthetic full for tape.

[EricB]

Hi foggy, correct!
Supposedly on a next version of Veeam support would come to create tape backups directly from NetApp SnapShots and/or SnapVaults (although i wouldnt know that would work, probably only increments to tape).

@rhaslund, thats true and another downside of using reverse increments to tape is that if you have a very large amount of data to be written to tape you have to manually calculate sufficient time for the tape job before the next source backup job would start again. (otherwise, because regular backups have a higher priority the tape job would fail because the source backup files will be modified again).
With forward this shouldnt be an issue as simply a new increment is written on disk without touching the previous restore point.

In this case i only need a monthly tape and its required to be a full.

Unfortunately I have no possibility to change the strategy (forward->reverse) in this current project but will keep it in mind for testing and future projects.
Thanks for the quick responses!

[veremin]

@rhaslund, thats true and another downside of using reverse increments to tape is that if you have a very large amount of data to be written to tape you have to manually calculate sufficient time for the tape job before the next source backup job would start again. (otherwise, because regular backups have a higher priority the tape job would fail because the source backup files will be modified again).

Just so you know - we're planning to address that issue in version 9.5, so that a source job can be automatically put on hold till a tape one is finished. Thanks.