Incremental Long Term Storage

[witbeckb]

We are evaluating why our offsite storage (backup copy jobs) are taking so much more than space than our old backup software, and we found that every consolidated backup (weekly, monthly, yearly) creates a new full backup file. In our old backup software the consolidated backups were still incrementals, with the oldest of the incrementals being a synthetic full that gets updated as it rotates past the retention date.

Is there a way to do something similar in veeam? We would rather have longer term storage over the risk of making the longer term retention based on an incremental chain (so one point being deleted could destroy the whole chain). Currently to get 4 weeklies plus 2 monthlies (for just 3 months of retention) is requiring more than 6x the data being backed up and adding a lot more than we anticipated to our offsite costs.

[nmdange]

There are two options to reduce the space usage. The best option would be to use REFS on Windows Server 2016 and leverage block cloning. Alternatively, you could use Data Deduplication, using either what comes with Windows Server, or a deduplicating appliance.

For 3 months, though, I think using incrementals would be fine. If you use block cloning or deduplication, the data blocks are still shared between multiple backup files, and thus any disk corruption with those blocks will corrupt all the files. So the only additional risk you have with incrementals is an administrator accidentally deleting the incremental backup file. So unless you are prepared to actually have totally independent full backup files with the required disk space, there is risk with data loss whether it's incrementals or block cloned/deduplicated synthetic full backups.

[witbeckb]

@nmdange - I should have mentioned this: the backups are going to an offsite 3rd party cloud connect repository, so I don't have control over the OS, file system, or storage appliance.

[foggy]

GFS restore points (weekly, monthly, etc.) are separate independent full backups. If you are ok with a longer chain that covers 3 months back, you can disable GFS and configure retention of 90 restore points, each of them will be incremental with only the very first being full.

[witbeckb]

@foggy - while that works, extending the production chain that far also wastes a lot of space. We don't need granularity down to the hour for a 3-month old backup, whereas a restore within the past week may need that granularity.

A potential fix is to add an option to the backup copy jobs that allows the consolidated recovery points to be a chain rather than independent full backups. This would have lower overhead and allow longer retention with the same hardware compared to independent full backups for each GFS restore point.

Does the new "archive tier" in 9.5U4 help with this at all? If not, could this type of incremental long-term storage be added as a feature request?

[foggy]

It is quite common that the monthly increment has almost similar size as the full backup, due to the amount of changes occurred inside VM during the month - so no gain in terms of space if it is a chain of increments. Your request is noted though.