Isn't virtualisation fundamentally about making one machine appear to be many, at least to exernal devices and systems? Other than the management and troubleshooting of the virtual environment itself, I'm struggling to see that many applications of exposing the "truth" via structured node addressing.
Structured MAC addressing might be of use, if it is necessary for each virtual host to have a different one. Using the locally assigned MAC address space would be the way to achieve that if it is important. If you need a multicast group just for the virtual hosts, you could just use conventional multicast methods, and then only have the virtual hosts join that multicast group. Alternatively, you could create a virtual network/link within the physical host, and then only have the virtual hosts be a member of that link. That would mean that link-local multicasts would only go to the virtual hosts. Making one of the virtual hosts an IPv6 router connected to the virtual link and the upstream "real" network would then provide access to the hosts sitting on this virtual network. On Wed, 26 Sep 2007 10:46:55 +1200 Brian E Carpenter <[EMAIL PROTECTED]> wrote: > I don't see why you would need to do that. If a subnet prefix > is a /64, that leaves 64 bits for virtual hosts on that subnet, > each of which could participate in neighbour discovery. Why do > you need structure in the interface ID? (Of course, you could > certainly choose to number the virtual hosts consecutively > within the IID space, but that's an implementation detail, > and might be undesirable for security reasons.) > > However - it is clear that massive server virtualisation > using global addresses is possible with IPv6. > > Regards > Brian Carpenter > University of Auckland > > > On 2007-09-26 08:36, Arul Kumar Chellappan wrote: > > Dear All, > > > > With virtualization gaining momentum, would there be an option in IPv6 where > > it has a mechanism to address a virtual host from the real world. Like the > > 128bit IPv6 address having [Network Prefix][Host ID][Virtual host ID]. For > > the backward compatibility, the host part could be subdivided into real host > > ID and the virtual host ID and range of virtual host IDs could represent a > > single real node. > > > > Yes, I understand about the major implementations now have no provitions for > > that but multicast addressed like ALL_VIRTUAL_NODE_ON_THIS_REAL_NODE that > > would help in easy management of that device. > > > > This is just my initial thought, with limited understanding on > > virtualization and IPv6. Could you please let me know your thoughts on > > this. > > > > Regards, > > Arul Kumar C > > -~- > > > > > > > > ------------------------------------------------------------------------ > > > > -------------------------------------------------------------------- > > IETF IPv6 working group mailing list > > [email protected] > > Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 > > -------------------------------------------------------------------- > > -------------------------------------------------------------------- > IETF IPv6 working group mailing list > [email protected] > Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 > -------------------------------------------------------------------- -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
