In message <878zdb4uor(_dot_)fsf(_at_)snark(_dot_)piermont(_dot_)com>, "Perry
E. Metzger" writes:
"J. Noel Chiappa" <jnc(_at_)ginger(_dot_)lcs(_dot_)mit(_dot_)edu> writes:
My own feeling is that we're just going to have to accept the notion
of our routers having millions of routes in them and go for algorithms
that scale better than distance vector or path vector so we don't
drive them into the ground while doing the computations. We can't get
rid of the desire to have huge numbers of routes so we have to find
ways to avoid nuking ourselves when we have huge numbers of routes. It
would be nice if we could come up with the perfect new architecture
but no one has yet designed it.
More to the point, no one even knows if such a system is designable.
If we stick with the current addressing model -- v4 or v6 -- we either
have to *enforce* topological addressing (with all that that implies
for the fate of multi-homing), or we have to hope for a fundamental
breakthrough in routing algorithm design. Lots of us would love to "go
for algorithms that scale better" -- but we don't know how to. It's a
research project, i.e., we don't know the answer and we don't know when
(or if) we will know it.
During the IPng directorate, there was a fair amount of interest in
designs that separated locators from addresses. It was argued
intensely, but in the end there was no consensus to move in that
direction. For lots of reasons, I think that that was the wrong
decision, but we're pretty much stuck with v6 as defined. The question
is how to move on from here -- and I hope that the road ahead works,
because we *really* need the extra address space.
--Steve Bellovin, http://www.research.att.com/~smb
Full text of "Firewalls" book now at http://www.wilyhacker.com