Re: Regionally aggregatable address space for multihoming

[RJ Atkinson <rja@inet.org>]

At 07:49 12/06/01, Iljitsch van Beijnum wrote:
>I guess this is true for the really large networks. Maybe this is 
>because they've lacked a good incentive to interconnect at more 
>locations. Geographic aggregation could be the necessary incentive. 
>They  are already present in many locations and local fiber is cheap, 
>so it could happen. On the other hand, it may not.

        Local fibre is cheap in some places, expensive in others.
In some places, long-haul connectivity is much less expensive
than local connectivity.  In other places, the inverse is
probably true.  When designing a global system, few generalisations
can be accurate.

        If we boil the above down, ISPs are incented to connect
over low-cost links, with regards to the distance of the link.
At the end, economics will dominate those business decisions.

        Given that, I don't see how trying to force a local
exchange system will be effective anytime soon.  And I don't
necessarily think that forcing such will actually provide
significant improvement.  I'm open minded about the latter,
but would need to see more substantial explanation of how
such actually improves the health of the global routing
system (something notionally similar to a paper in ACM CCR
describing the reasons for improvement in global routing system
health would be a nice start :-).

Cheers,

Ran