On the one hand, currently building out some infrastructure, I can see the appeal of this. On the other hand putting another layer between the physical infrastructure and the actual infrastructure seems fraught with risk.
The risk is that generally when something is 'broken' (and by broken I mean say the port on your 48 port switch has decided to auto negotiate 100Mbit always), virtualization systems like these result in 'poor behavior' but not clearly broken behavior in a system. That makes fast diagnosis hard, fast detection hard, and isolation even harder.
The risk is that generally when something is 'broken' (and by broken I mean say the port on your 48 port switch has decided to auto negotiate 100Mbit always), virtualization systems like these result in 'poor behavior' but not clearly broken behavior in a system. That makes fast diagnosis hard, fast detection hard, and isolation even harder.