I have read conflicting information on best practice recommendation for speed/duplex settings for 10GbE interfaces. In this case, we will also be passing FCoE traffic.
(1) Is there a compelling reason to choose one over the other forced 10000 Full or Auto Negotiate?
(2) All VMs show 10GbE interfaces as expected. A colleague of mine spoke of an issue he had with a VM that started a storm of traffic with another VM on the same Host and vSwitch resulting in a Host CPU spike. He toyed with the idea of tweaking speed/duplex settings at the VM level. I advised against that but, it got me thinking of ways to mitigate such events. QoS at the switch layer the only thing I could think of. Does anyone else use other preventative or restrictive measures to guard against this?
