CSV on Hyper-V 2016

[Tos]

Hi

We have question about backup of VMs on CSV on Hyper-V 2016.

#1
Can Veeam backup VMs on CSV on Hyper-V 2016 ?
There was description of Hyper-V 2012 R2 and Earlier in the below document.
https://helpcenter.veeam.com/docs/backu ... tml?ver=95
But, there was no description about the CSV of Hyper-V 2016.

#2
If #1 yes, do we need to use Hadware VSS Provider ?
In Hyper-V 2012 earlier, even if we backup CSV, we think Veeam can use Hardware VSS Provider or Software VSS Provider.

Regards
Toshi

[Mike Resseler]

Hi Toshihiro,

It is not a problem to protect VMs that reside on a CSV on Hyper-V 2016. You still can use a Hardware VSS provider but your hardware needs to support windows server 2016 and not many hardware devices today actually support it. The reason why it is not in the same link is because backup in ws2016 is completely different (and new) compared to previous versions.

You can use the hardware VSS (off-host proxy), but you don't need it. If you have a balanced Hyper-V cluster and hosts have resources available, then using a software VSS provider will work also fine (on-host proxy)

Cheers
Mike

[Tos]

Hi Mike

Thank you for reply.

The advantage of using a hardware provider is that you can use an off-host proxy ?
Is there no benefit like Backup from Storage Snapshots of VMware ?
For example. the speeds up backup operations and reduces the impact of snapshot removal on the production environment.

Regards
Toshi

[Mike Resseler]

Hi Toshi,

No, it is not the same. Both techniques are different. The advantage you have with off-host proxy is that the calculations and heavy work are done on the off-host proxy which means you need less resources from the Hyper-V host itself (on-host proxy). But the snapshot removal remains the same. However, since 2016, and the fact that you can run those VMs on ReFS makes another big difference. ReFS and block cloning technique allows that merge (snapshot removal) to happen much faster and with less IO needed. So certainly something to investigate while you are creating your cluster (especially when you combine it with S2D or classic storage spaces)

Makes sense?