Switching from reverse incremental to incremental, how to retain full backup to tape?

[cerberus]

Hello,

We use reverse incremental backup mode today, to back up our VMware cluster, and all the goodies are enabled (CBT, REFS for fast clone, and we backup from Pure storage snapshots), and then write the data to LTO9 FC tape, daily B2D2T (full daily backup on disk, and on tape), no issues. Works great for many years and is very optimized for a "slow reverse incremental backup mode" B2D job (30min backup job at 2.2GB/s and a 10-hour tape job at about 600MB/s).

With Veeam removing reverse incremental in future versions, and while I dislike having to mess around a well-established and time tested backup setup, we have to start looking at the other backup mode options, my main requirement is to ensure the latest backup on disk and on tape is always the full backup, particularly on tape, specifically I do not want to have any incremental chains at play if/when I restore backup from tape.

Based on this tape-f29/always-full-backup-for-tapes-t84711.html, HannesK goes on to say that the options available are:

• Use forward incremental forever backup chain without synthetic / active full. then use a simple media pool and synthesize full every day

• Use forward incremental with daily synthetic full (for disk space consumption reasons, only do that on REFS / XFS)

Now I don't mind that if on disk, the recovery points use a backup chain, where the latest backup is not a full backup, but on tape I have a requirement where it needs to be a full backup (synthesize full every day at the media pool level).

Looking at the other option, where we enable daily synthetic full on the backup job, if the underlying OS is Windows Server 2022 and the data disk is REFS, what sort of storage disk space usage can one expect if running daily synthetic full mode? Veeam describes synthetic full as...

When you perform a synthetic full backup, Veeam Backup & Replication does not retrieve VM data from the source datastore. Instead, it synthesizes a full backup from data you already have in the backup repository. Veeam Backup & Replication accesses the previous full backup file and a chain of subsequent incremental backup files on the backup repository, consolidates VM data from these files and writes consolidated data into a new full backup file. As a result, the created synthetic full backup file contains the same data you have if you create an active full backup.

However, with REFS at play, if my initial B2D job is 20TB, surely the daily synthetics aren't going to be another +20GB per run?

Any pros/cons going with "virtual full tape" vs. "daily synthetic full"; which one is better in terms of performance?

[david.domask]

Hi cerebus,

I think what's missing in your understanding is the space savings from Fast Clone which creates "spaceless" Synthetic Full backups. So only the space for an increment is _required_, if a block is eligible to be fast cloned, it will be fast clone'd as opposed to copying the block into the Synthetic Full backup.

Either solution above works; Virtual Fulls with Tape are in most cases the fastest way to put a full backup on tape, and you could do this with the first solution Hannes suggested (Forever Forward Incremental for your source jobs, schedule Virtual Fulls as desired). I think this is likely best bet for most environments.

Also, regarding the use-case of daily fulls to tape with Reverse Incremental, this use case is not lost on us and there are plans to address this "soon to be lost" use case and ensure daily fulls on tape, so stay tuned; either solution above is fine for now however and will achieve the same goal as Reverse Incremental + daily fulls to Tape.

[cerberus]

I ended up going with the Forever Forward Incremental option, while enabling daily virtual full to tape, works great