vPower for Hyper-V

[J1mbo]

It's easy to understand vPower on vmware since it just uses NFS; I'm interested how this works with Hyper-V.

Hence looking for any deep dives or similar on the subject.

Many thanks!

[Gostev]

Have you read the User Guide? Pretty "deep dive" explanation there.

[J1mbo]

Thanks, the 6.1 documentation is well hidden! But here it is:

3. On the backup repository and on the destination host, Veeam Backup & Replication
deploys a pair of agents that are used to mount the VM disks from the backup file to the
dummy VM.
4. On the destination host, Veeam Backup & Replication starts a proprietary Veeam driver.
The driver redirects requests to the file system of the recovered VM (for example, when a
user accesses some application) and reads necessary data from the backup file on the
backup repository via the pair of agents which maintain the disk mount.

With Server 2012 I understand file level storage will be supported in much the same way that NFS is for vSphere, but using SMB 3 - presumably that will streamline this process?

[J1mbo]

BTW, it would be useful for internal documentation if there was a permanent link to user manuals and similar, for example http://www.veeam.com/manuals or something like that. The links presented look version specified at the moment.

[Gostev]

Why would we do this? There are no benefits to this approach. We really like the current approach, because it is significantly faster. We only use shared storage approach with VMware, because with ESXi has no parent partition where we could run our own code.

Permanent link to manuals:
http://www.veeam.com/vmware-esx-backup/resources.html

[J1mbo]

Why would we do this?

Because all complexity is bad! An extra agent - however simple - always carries with it risk of something going wrong or conflicting particularly with some future update; which to my mind is precisely what we want to avoid within the hypervisor layer.

Anyway, I'm interested in the assertion that the Veeam agent would be faster than running an SMB share. Surely the underlying storage latency - typically in the ms range - is far more significant than any processing overhead at the host?

Permanent link to manuals: http://www.veeam.com/vmware-esx-backup/resources.html

Thanks for this!