In the early morning and occasionally during the working day we are experiencing lots of 4 minute freezes in our Remote Desktop Session Hosts (Windows Server 2016) running in Azure. The servers are quite big as they are serving Windows 10 like general desktop sessions to 30+ users.
Only some of RDSH VMs are affected on each occasion. Different hosts on different occasions. One host may be hit twice in a single period. A period of freezes may be spread out up to an hour. There are more VMs affected in the early morning after they all power on than during the day.
I have a background script running on the VMs watching for Audio service hang ups and its logging indicates the 4 minute freezes involve the CPU pausing. Essentially if the CPU wasn't pausing the script would be logging a message after just 1 minute. My guess is the Azure Hyper-V hosts are automatically rebalancing on changes in load and to do so are migrating VMs between hosts. As our VMs use a lot of memory I believe the 4 minutes is mostly the time taken to copy the RAM contents to the other Hyper-V host.
Is this a reasonable conclusion? Are there other possible causes?
Has anyone managed to prevent their VMs from being moved?