Gateway servers per job?

[ejenner]

Was just setting up some new jobs and the console promoted that I'd not specified a gateway server for the repository. Which was true.

So I looked to see which hyper-V host was storing the VM I was targeting and thought I'd set that as the gateway server.

Then I realised as I have more than one hyper-V host in this particular non-clustered environment setting the gateway server for the whole repository to the host local for some of the VMs wouldn't be ideal. i.e. when the other VMs from other hosts started to backup they would use the first Hyper-V server as their gateway.

I don't know the full details of how gateway servers are implemented within the software but it seems a bit of a limitation if you can't set the gateway server depending on the VMs location. It does generate a warning message when you don't have a gateway specified so clearly there is some disadvantage to having an inappropriate gateway set and logically it seems pretty obvious that would be the case.

So could the gateway selection be in the job settings in addition to the repository settings or instead of the repository settings?

[veremin]

If selected repository cannot run target data mover, then you need to have a gateway server for the repository - intermediate component through which the communication with repository will be established. You can set there a specific server or set automatic mode and let backup server choose appropriate gateway server automatically.

For Hyper-V backup jobs "automatic selection" almost always means that source Hyper-V host will be chosen (exactly what you are after).

Thanks!

[ejenner]

Will the same apply for any job or is it just lucky that in this instance we're talking about stand-alone hyper-V?

[veremin]

No, you are not lucky that you have only one host, all backup jobs will try to use their source proxy (Hyper-V host) as a gateway server, which means it will differ from one job to another in case of multiple Hyper-V hosts.

I'd also recommend reviewing the referenced documentation, should help to grasp the concepts better.

Thanks!

[ejenner]

Not what I meant. If the job was an agent job or a VMWare job or a File Copy Job or something different to this scenario. i.e. is automatic selection a perfect substitute for being able to chose manually from the job settings?

[foggy]

Automatic is always preferred unless you have specific reasons to have manually selected gateways.

[ejenner]

Correct. And if you have a specific reason to select a particular gateway doesn't it make sense to be able to set it per job instead of per-repo?

[foggy]

Gateway selection is typically an infrastructure-related choice rather than a backup workload-related. But you always can set up a separate repository pointing to a different folder in the same storage/location for some jobs that require a different gateway.