Dedup not working as expected

[marc_al2]

Hello,

I have made some tests and I think that the dedup doesn't work as expected.
I have made a full drive backup. After that I have made a copy of a VHD in a directory (5 GB).
The new backup took 4 GB (more).
I made a copy of the copy, it took 4 GB (again) to 8 GB for the two backups).(so both files are exactly the same)
I made a copy of the two copies (so I now have 4 tiles the same files). The new backup adds 8 GB for a total of 16 GB (so the dedup doesn't work too within one run).
Am I missing something or there is no dedup?

Thank you
Marc

[Vitaliy S.]

Hello Marc,

Dedupe is done within the restore point, not the entire backup chain. This is done to avoid issues when, for example, old restore point is removed by a retention policy and data which is shared between other points gets removed as well.

P.S. Please correct me if I'm wrong, but incremental backups usually have unique data that cannot be deduped with a previous restore point, so I assume this should not affect jobs in real environment, right?

Thank you!

[marc_al2]

Hello,
Thank you for the explanation.
I had the idea that it was working like duplicati or crashplan. (in Duplicati, every block that has to be deleted is kept if it is used by dedup, and when it want to retrieve some space (if it will gain a specific percent), it does a new copy of the blocks that are still in use).
The case where it would be helpful would be (for example) with virtual machines and updates...
Another example (I had) was when only a part of a file is modified in file mode, you must copy the whole file again while there are only a few megabytes modified.

Marc

[Vitaliy S.]

The case where it would be helpful would be (for example) with virtual machines and updates...

Sounds like a perfect use case for our flag-ship product Veeam B&R that does backups of VMs

Another example (I had) was when only a part of a file is modified in file mode, you must copy the whole file again while there are only a few megabytes modified.

Yes, that's a good point, but if you want to have efficient global deduplication one of the ways to achieve that would be to use Windows Server 2012 R2 with deduplication enabled as a target repository for your endpoint backups.

Thanks for the feedback!

[Dima P.]

I have made a full drive backup. After that I have made a copy of a VHD in a directory (5 GB).

By the way - backup location is excluded from the backup by default, so every restore point produced by the backup job would not affect the overall size of the backup files.