is repo share blocks between VMs ?
Of course, should they have some in common. That is more likely to happen on subsequent backups of the same VM or among VMs comming from the same template.
while testing, backup 3 windows server 2016 , same windows update level. total size was equal to the total size of 1 repo per VM.
repo which share multi VM, is only prune operation dedup the blocks ?
We aren't fully getting you. If you are using some translator, please try to use shorter phrases addressing one concept at a time.
The amount of deduplicated blocks depends on many things. Per instance if the three servers are not block aligned exactly the same to the VMFS file system, you will not achieve any crossed deduplication. That doesn't really matter though. (c)XSIBackup has been designed to deduplicate successive backups of the same data set (namely the same set of VMs), thus the deduplication you achieve is mostly gained through backing up the same set, not by finding the same blocks in different VMs, which is rare.
No, you don't need to prune to achieve deduplication.
sorry for my poor english (french here ...)
thanks I get it : dedup works on aligned blocks , so even with my same windows "base" , dedup cannot be done at that level.
incremental backup per vm works really nice , speed & delta size.
so 1 repo per vm is prefered for my head
Even though deduplication is more likely to happen on the same VM for obvious reasons, it is quite common in datacenters to deploy VMs from templates, thus, any VM that belongs to the same deployed template will still be highly likely to be sharing a good amount of blocks with some other VMs. In case of big datacenters this will represent a huge amount of data.
It is up to you and your knowledge of your own data to determine and/or test whether you will achieve crossed deduplication between VMs or just serial deduplication among the same set's succesive backups.