With the current situation, we have to take a snapshot and deploy new WVD session host servers whenever there is an application upgrade or you need to manually update all the session host servers or use other tools like SCCM or Invanti etc... but my question is there a way to achieve this without using any other tools because in Citrix we have the machine creation service inbuilt to handle this. So that we can push an application update to all session host servers simply updating the master image.

We have a large customer deployment around 100+ WVD session host servers. in such a scenario what is the best and easiest way to handle the custom image management?

Using Azure DevTest labs as inspiration and the power off scheduler already available to Azure. I believe that a host pool should have a native power on and power off policy with a scheduler that can be mapped to a time zone, with the (optional) ability for desktop user to extend by 2 hours/skip the day of power management.
My story is that I have a pool of dedicated machines that I will be providing to developers. We have an expected hours of operation of 6am to 6pm Mon-Fri but there will be sometimes when someone needs to stay back or work on the weekend. Having
a) the native ability not to have a dedicated machine to power on machines or need to individually set power off policies
b) the ability like Azure DevTest labs for an optional extend/opt out email sent to a user
would be extremely valuable and enhance the product greatly.
Example user ability opt out with Azure DevTest Labs
https://docs.microsoft.com/en-us/azure/lab-services/devtest-lab-auto-shutdown#set-auto-shutdown-policy-for-a-lab

I think we need the ability to migrate active user connected sessions (not VM) from one host to another, similar to live migration VMs between hosts in virtualization. The objective is to optimize pool utilization, similar to optimize Hyper-V cluster utilization. So when host utilization is low, consolidating active sessions into less hosts and Stop the rest would be very cost effective.

We only have a rather small pool with 10+ hosts that serve 200+ CCU and by end of the day can hardly find any empty host even with DepthFirst enabled and the low utilizing hosts are set in Drain Mode. I can imagine the admins of bigger pool, like 100+ hosts or so, pulling their hairs desperately trying to consolidate the pool without disrupting end users workload.

Previously it was impossible to have such “live migration” feature. But today we have FSLogix and soon will be MSIX, which detach both user profiles and apps from the VM itself. Hopefully it could be possible to have such feature developed soon.

Power Off Personal VM's initiated by teh end user. This is the reverse of Start On Connect.
Provide the end user a way to deallocate their own Personal Desktop. This means they can power it off at a time that is suitable for them, rather than having it automated using another tool, which may power it off to early if working late or at weekends, or leave it running incurring costs if finishing early for the day. It also negates the need for IT to be involved in this. The end user will know that if they have powered it off, at 5 PM that they aren't incurring compute costs when its deallocated, plus they know that in the morning of the next day it will power on when they need it.

HI there,

Currently it seems that for Azure WVD - Microsoft only publish the single host /32 IP's for its WVD endpoints.

Whilst I am not really a fan of single host especially in public cloud, would it be possible that they can bring this inline with what they do with O355 services and allow based on a summarised subnet basis ?

this makes like a lot easier (IE

-not such a huge object list in firewalls
- Cleaner solution
- A lot less risk in terms of microsoft changing and causing an outage on the platform whilst we look at scripts and re-inject objects in to the address books as your likely to change to a possible IP in the same subnet.

• Although we have some newer kit on our estate, we still have some older firewall infrastructure awaiting refresh, this causes issues when we are nearing some of our object and policy counts, and ideally we dont want huge lists of indivual IP's floading around the address book




• there are more benifits to doing this in a subnet based format

I assume you have had a lot of interest in this feature from other entities, could this be pushed forward as an upcoming feature release ?

Thanks