I have heard and read a lot about the reccomendation to seperate the page file onto a different disk in a virtualized environment, in order to increase performance and save bandwidth.
From what I have seen, heard, and read from vendors (NetApp, etc.), all this was always mentioned in relation to VMware. No where did I see anything of this kind mentioned when people were talking about Hyper-V. Did I maybe miss something written somewhere? Is there a reason?
Separating the page file on any hypervisor will increase performance on any array, but not a lot. On a NetApp Controller It will also help maintain better dedupe savings, better dedupe will lead to better performance… However, those recommendations are based purely on performance, and don't take into account the operations impact. In this case it's my opinion that the minimal performance benefit doesn’t justify the complexity it introduces. I’d keep it all on one drive.
My concern about not using page file separation is the addition space consumed by snapshots when replicating VM's using SnapMirror. Separating the page file can reduce the amount of data replicated across WAN links. The default settings on a single guest can be about greater than 10GB.
If you are not replicating the VM's across a WAN, I agree that the complexity of maintaining multiple VHD's outweighs the benefits.