I've just discovered the requirement for our B&R server to be on 2012 with De-dupe enabled if we need to recover a 2013/de-dupe-enabled VM.
This kind of forces our hand into the split model, with a B&R VM and physical server for repositories etc. (our current all-purpose physical B&R unfortunately can't be upgraded to 2012 due to lack of HP support).
I was kind of warming to the change as it then allows us to better protect the B&R server but something that just struck me is a question on how IR vms are mounted.
Currently, with everything physical (and a few Proxy VMs to aid concurrency) if I IR a VM, then the physical server mounts the backup files directly from either local storage (ok-ish performance) or FC storage (great performance) depending on where that particular VM backup currently sits. However, if the B&R server is moved to a VM, and the physical is "demoted" to proxy/repository tasks, would that mean that if we do an Instant Recover, that the B&R VM would mount everything (via the network instead of via FC/DA storage) with all the resulting performance hits, or would the mounted VM/backup still be via NFS on the repository server so no different to how it would currently work?
As I type this, the later seems the most plausible but I wanted to make sure rather than make assumptions!
[I've used IR numerous times before with great results, just wanting to make sure that this process isn't significantly impacted by the architecture change]
[New Sig: PLEASE get GFS tape support for incrementals!!!]