Hi everybody,

I am having a pretty annoying problem with our Hyper-V Cluster. One, and only one of our VMs keeps getting into a state where disk performance is absolutely horrible (a few tens of KBps on sequential transfers). The interactive performance in a RDP is also affected.

However, as soon as I do a Live Migration to the second node in the cluster, the performance goes back up to where it should be. Live Migrating it back to the original cluster node keeps the performance up, that is until a few hours later, at most a couple of days, the performance goes back to being horrible. Apply, rinse, repeat.

In the parent partition on that cluster node, the disk performance is just fine, irregardless of what happens in the VM. No abnormal CPU usage to be seen, either in the parent, or in the VM. No particular events logged either. The Live Migration is also pretty slow when the VM is in the affected state, moving it back performs observably faster (i.e., once the problem is "reset" through live migrating it on the second node).

The cluster is WS2012, the VM is also WS2012. It only functions as a file server, no other software installed, not even an antivirus. The VM is however relatively large (4 vCPUs, 32 GB RAM, a 120 GB dynamic VHDx as a system drive, and a 24 TB dynamic VHDx as storage for the file share).

There are a few mentions online about similar problems, but no concrete tips on what could cause it. Anybody have any ideas?

Hi Amyd80,

"It only functions as a file server"

"120 GB dynamic VHDx"

Based on my knowledge , file server with dynamic VHD will bring critical fragment issue .

Please try to backup this VM then convert the dynamic Vdisk to fixed Vdisk .

Verifying whether the performance is really changed .

Hope this helps

Best Regards

Elton Ji