Advice on design with Hyper-V cluster backups

[Blue407]

Hey All

We have a Dell VRTX server, its quite unique in that it has 4 x server blades plus storage built in that's available to all the blades (Built in shared storage)
Its currently running Server 2012 R2 with a hyper-v cluster. (Ongoing project to upgrade to Server 2016)

Currently we have a dedicated separate backup server, linked to each of the blades with 10G Ethernet.
Veeam 9.5 is running on the backup server and it backs up to on-board disks within this server. We have both backup and hyper-v VM replication jobs running to the backup server.

This all largely works okay, but I'm looking to improve performance.

I'm considering the idea of using one of the 4 blades as a dedicated backup server in the first instance. Adding a PCI based SSD, or pair of SSD drives which will appear as local storage to the blade.

Can Veeam use this very high speed local storage as a temporary cache, so for each job it backs up the running VM into local SSD storage and then copies this to the backup server repository?
The backup blade would see the VM's as local shared storage so would have fast access to them.
The advantage I see is that the backup blade sees the shared local storage, it has very fast SSD for caching so the backup job finishes very fast. It doesn't then matter if it takes a little longer to copy to the dedicated backup server.

I hope I've explained this properly and look forward to suggestions and ideas
I think our infrastructure is fairly unique as most hyper-v clusters use SAN for shared storage. I am therefore looking to take advantage of the share storage capability to improve my backups.

[Mike Resseler]

It's rather difficult to give you the correct answer since I don't know that type of hardware and the capabilities. However, you are probably using Hyper-V in onhost mode so most resource intensive work is already done on the Hyper-V hosts itself. The other work is being done over the network. So from my point of view I see no advantage at all. Unless I misunderstand the setup you are trying to create

[Blue407]

Fair comment. So, the host that is currently running the VM does all the work as its currently the owner? Even though other hosts have access to the VM files, they do not currently have ownership and can't therefore back it up?

[Mike Resseler]

Paul,
I didn't explain it well enough. Sorry about that. Onhost means that the heavy lifting is done by 1 off the Hyper-V hosts in your cluster. Not specific the owner. So it could be any host at that point in time depending on resources available. Especially with 2016, you will see that it makes a huge difference. Previous versions of Hyper-V could have been an issue since it would be a backup in redirected mode but with the latest backup framework, that has changed

Hope I explained it better now

[Blue407]

That makes much more sense, thank you.

So once all hosts are 2016 and the Veeam agent is installed on all of them, backup and replication jobs will automatically share the jobs between the hosts? not necessarily the host that is currently running the VM?

[Mike Resseler]

Yes, unless you specific a specific host that always does the heavy lifting...