We are running a 4 node 2012 R2 Datacenter Hyper-V cluster.
SAN is configured as CSV.
Last Friday and a couple times on the past, we have received alerts that our platforms are crawling. Log in, application availability, scom performance alerts etc... When checking the SAN, we were able to see massive Reads from the ISCSI initiators.
Nodes are responding fine.
VMs are slow
The Iscsi connections are reading upwards of 800 Mbps and writing 200 Mbps.
We attempted to live migrate the VMS over, but that had issues (timeouts) due to the iscsi throughput.
Question : Is there any powershell / perfmon KPI I could used to isolate what VM(s) are doing the massive reads through the Virtual Switch?
Your best bet is to use Perfmon both in host and guest systems. There is a free tool, PAL, that can provide you with a pre-determied set of Permon counters. It will create you nice HTML reports of the counter logs too.