What's best practice when dealing with Azure regions in a DaaS config? When we initially setup a hosting connection in Azure we created a resource group for it with Cloud connectors, as required. But now we are building out additional workloads in a different Azure region. The hosting connection is still the same and the new region appears as an additional resource under it. Are we required to build out a new resource location in DaaS for it, which then automatically creates a separate zone? It seems like everything could just live under the existing Resource location / Zone.

Yes you should have a separate RL for a separate region in Azure otherwise things will not behave as you would expect when you have a failure. IN DaaS since its a single site you would configure failover using zone preference. So if you want all you users to use say azure east 2 normally and failover to azure south central if east 2 is not available you create a group wiht your users and put it in the east 2 zone then configure your resources to prefer the home zone. Then put catalogs from easch zone in teh same delivery group. If you want to spread users acriss both just doine use the home zone.

I have always created separate resource locations with cloud connectors for different regions in azure

A new RL with new CCs was how we built out the new region. But then MCS just wants to point the VMs at the existing CCs in the first region, and the DaaS console treats the new region as a subfolder of the main one. Hence my confusion.

Thats because you do want to create a separate hosting connection and put that in the new zone. I would also use a different service principal for it to lessen the risk of issues if you have a problem with the service principal.

A different hosting connection for each region? Interesting.

Have you tried to force the mcs catalog to look at the new resource location only. Set-brokercatalog "name" -zoneuid "newzone". I recently did for mine and works with no issues

I always turned off auto discovery of connectors for mcs deployments and forced VDA registration via policy. Auto discover does some weird things and often screwed up appropriate registration points