Joe,
| Well, terms like "overlay" and "virtual network" are admittedly used | by many people with different meanings. It would be useful to clarify; they are the same to me.
To me, the distinction relates to whether the topology is supported/hosted directly in the infrastructure or not. In any case, whatever the terms, I think there is an interesting distinction to make about whether the underlying infrastructure (or substrate or whatever) supports the virtual networks, as it creates additional requirements for the network elements.
VM is the result; MV is the process by which VM is supported. I don't see how either requires sharing of physical memory, as per the example I gave previously. The key aspect is the additional layer of indirection of the address space.
That's indeed the key aspect of virtual memory. That said, I think it's useful to think about the notion of a shared physical memory supporting the abstraction of multiple virtual memories. To me, that's what a term like "memory virtualization" connotes. I agree that virtual memory doesn't necessarily require the sharing of physical memory but that, when such sharing does occur, it does imply some set of mechanisms and policies to enable such sharing.
I wouldn't either, but I thought the question was to define what "network virtualization" means, not what a "virtual network" is. The underlying substrate that enables the sharing of the physical network is, to me at least, what the term "network virtualization" means.Mechanisms that support network virtualization can be interesting, but I wouldn't either explain or motivate VNs by their supporting mechanisms;
-- Jen
Note Well: Messages sent to this mailing list are the opinions of the senders and do not imply endorsement by the IETF.