When snapshotting a running VM, I notice that the affected VM's performance is degrading heavily (leading to bizarre response times in the magnitude of minutes where similar actions might only take 2-3 seconds when the VM is running normally) during the first 50% or so of the snapshot creation procedure. The snapshot procedure might take a while with memory-intensive VMs as it is heavily bottlenecked (presumably by the the compression of the memory's data which happens before it is written to disk).
Most of the on-the-net documentation is mainly dealing with the vhd/vhdx-related aspects of a snapshot and the performance implications it might bring. But I am more interested in the part where the memory dump is written to disk - what is happening in this process and how would I even try to determine the duration of a snapshot-related impact?
