> From: Lixia Zhang <[email protected]>
> This proposal was considered and debated at the time, but did not
> fly ... mainly due to the reason that has been articulated in this
> thread of exchanges: the model does not match the ISP economics.
To expand on this a tiny bit, let me give it to you as Yakov concisely put
it to me:
In routing (i.e. path selection), for the overhead of routing to scale
'reasonably' in large networks, either the 'addressing' (i.e. the names used
by path selection) has to follow the topology, or the topology has to follow
the addressing; i.e. there has to be a certain level of congruence between the
two.
(And by 'topology' I mean actual network connectivity - we're mis-using
the term 'topology', but it's probably too late to fix that now, although
if someone has a better alternative term, I will cheerfully switch to
using it... :-)
Unfortunately for geo-addressing, for economic reasons, topology usually
follows traffic patterns (i.e. you lay/light-up fiber to match traffic flows),
and traffic patterns follow relationships in the real world (e.g. between
customers and their suppliers/providers). So driving the topology from the
addressing end is not possible.
Instead, the only way that works is relationships -> traffic patterns ->
topology -> addressing.
Noel
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg
