CBT vs Proprietary Filtering for Windows Deduplication

[veeeammeupscotty]

I understand that Windows server deduplication wreaks havoc on CBT due to the various optimization jobs that can cause a large amount of changed blocks even for an incremental backup, but I'm wondering if Veeam's "proprietary filtering" would allow for smaller incremental backups in this scenario. Even if so, I realize there are probably other drawbacks to not using CBT such as more disk I/O/, CPU load, longer backup times, etc, but even with fast clone, we keep running out of space due to massive CBT incrementals.

[csydas]

Hi JP,

Not really. The issue with Dedupe and big increments is the C and B in CBT, as you pointed out. Even the proprietary method is still just looking at the blocks, not the actual content within the data block.

That's the big trade off with dedupe on a VM - you can save space on the datastore, or you can save space on the backup storage. Avoiding CBT is just going to be longer reads as you have to read the whole disk without any benefit from - Veeam is still going to see changed blocks and move those blocks to the backup.

Unfortunately it's just something you need to plan for; treat dedupe VMs as highly transactional VMs and plan the backup space around it. For our fileserver, we ended up just doing periodic fulls because it math-ed out to be less space over all.