RE: Architectural approaches to multi6

["Dmitri Krioukov" <dima@krioukov.net>]

Thanks for replies so far.

I'd just like to note that I used the term
"routing" in a very narrow sense. To be
precise, I meant the "routing function."
The definition can be found, for example,
in: http://citeseer.nj.nec.com/gavoille01routing.html

For example, separation of "locators" and
"identificators", and then using various
dynamic binding mechanisms, is not routing
in this sense.

With these definitions, I guess more
people would agree with my initial
assertion since the multihomer is
necessarily a node on the graph and
the shortest path routing was found
to be incompressible on generic graphs
(see the above link but note that it
deals with the ideal world (static
and centralized), yet it provides
valuable information on the fundamental
limitations of any routing (even
not yet invented)).
--
dima.