Re: Aggregating channels...

[Joe Abley <jabley@automagic.org>]

On Tue, Dec 18, 2001 at 08:57:57AM -0700, dreamwvr@dreamwvr.com wrote:
>   Basically not sure what your referring to as the right path.
> However it seems to me that a variation on telephony TDM 
> would work theoretically. Mapped to inet characteristics 
> rather than telephony characteristics. 

Circuit-swiched networks incur a call setup and teardown overhead
once per call. Packet-switched networks need to make routing
decisions once per packet. These are fundamental differences.

If you're interested in how the routing system might be adapted
to scale better (or in whether the routing system needs to be
adapted, since the scaling constraints are not known) go look
at the ptomaine IETF working group and at the numerous presentations
on this topic at the last few NANOG meetings.

If you're interested in other approaches for multi-homing, which
perhaps do not involve the routing system to carry the associated
state, go look at the multi6 IETF working group.


Joe