I use dedicated VM's as Veeam Proxy's across our infrastructure.. And since backups and replication only happen at night, these VM's sit idle all day long eating CPU cycles, ram and disk i/o on the hosts, I know an idle VM hardly consumes many resources, but it's still something and sometimes every bit counts. So my thought was it would be cool if VBR could hibernate the proxy VM after a backup/replication job completed and power it back on before the job starts again so the VM isnt wasting resources during the day.
You could always look to script some of this yourself ( depending on the number of jobs you run, a post job script could shut the proxies down and you could start your backups with a script to power them back on ) , but remember that there are times outside of the backup window you'll need those proxies ( copy jobs / restores / surebackup ? ) - when are you going to maintain those proxies as well ? Are you heavily constrained for host resources at the moment?
Hey Chris, Definitely good idea to toss in a script to take care of the hibernation. You bring up a good point about needing the proxies for other functions, I think this is an even better reason to have to feature integrated in to Veeam. I do currently have a VDI cluster that is very CPU constrained, it's over committed and no new budget for replacement/upgrades for the servers any time soon .. Trying to save resources on this cluster in particular is what made me think of the idea in the first place.
As far as maintenance on the servers, they would just be put on a schedule in change management and fired up for maintenance when needed.
I'm guessing that using an alternative off-cluster proxy isn't an option - what storage is your VDi cluster using ? perhaps a direct san proxy would free up a bit of load on it?
you could always put the proxies into a resource pool with a low amount of shares , all of your other workloads will be prioritised over them , thus reducing the idling impact even further?
Is your backup server a virtual machine? If so, you can leave it as a default proxy for any unintended activity happening during a day, and power other proxies closer to a night via PowerCLI or more specifically these commandlets: Start-VM, Stop-VM. Thanks.
If the virtualized environment is too overloaded I would go for physical proxies, but also check for its general state. I don't know of how many proxies we are talking about, but if some VMs are creating all those problems, even without them the load on the cluster sounds to be high anyway.
Luca Dell'Oca Principal EMEA Cloud Architect @ Veeam Software
.. Trying to save resources on this cluster in particular is what made me think of the idea in the first place.