-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 PS - some other details below; I had neglected to address some of the significant concerns about separable routing tables...
Pekka Nikander wrote: .... > I am framing my argumentation below as follows: > > 1. There are architectural alternatives to today's virtual memory. > 2. Virtual machine is more than virtual memory. > 3. None of our current tunnelling practises come even close to the > level of abstraction provided by either the virtual memory or virtual > machine abstractions. > 4. Even if we may get a clean virtualisation architecture with IPv6, we > are not there yet, and may not be clear where, exactly, we should be > going. > ... > > 2. Virtual machine is more than virtual memory > ============================================== > > The majority of today's hardware does not support full virtual machine > functionality, i.e., virtualising also the protected mode. In the main > stream architectures it has been appearing only recently. > Consequently, many of today's virtual machine products either use > software to "cheat", resulting in architecturally ugly (but useful!) > solutions, or are fully implemented in software, with the apparent > performance penalty. > > My point here is full virtualisation, such as in the case of virtual > machine, requires much more than just virtualising the address space. In the process of developing the X-Bone architecture, we recognized the need to keep spaces separate, to the point of needing support for different clusters of interfaces. Each cluster needs to be assigned to a forwarding table, where addresses could be unique within the virtual net each cluster is attached to, but reused across clusters. Marco Zec implemented "clonable stacks," which turned out to have the functionality we needed for this part of the endsystem virtual network support. This includes the notion that "inaddr-any" is meaningful only within each cluster - i.e., within each cluster there is a "weak host model" and between clusters there's the "strong host model" (both from RFC1122). There are other aspects of support needed. One includes file system virtualization to allow multiple programs to run with different 'root file systems'; this is key to revisitation, the use of a single physical node more than one times in a single overlay (akin to VM paging). Another includes the need for two-layer tunnels, to virtualize the link layer as well as the network layer. There are others, described as noted before in some of our groups' papers. ... > Should the routing system better support local-only routes? This is, FWIW, the kind of support that's already available in Zec's clonable stacks, and it is (IMO) required. It would also be useful to be able to install local forwarding algorithms and entire local stacks, but that's not currently available (though precursors have been tried, e.g., in Active Nets, our own DataRouter, etc.) Joe -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (MingW32) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC+j97E5f5cImnZrsRAjrWAJ9AzHBPr/joDQ2RwteLjKhKrEaU7QCeLMQu 0a5HKbBLyeSoaHxsQrBzobk= =TYAN -----END PGP SIGNATURE----- _______________________________________________ Int-area mailing list [email protected] https://www1.ietf.org/mailman/listinfo/int-area
