Does that constraint remains if peering happens closer to the edge to AS
that are more regional in nature. For instance, say the FCC mandated
peering at the LATA level over IP-based IMT's/Bill'n'Keep trunks.
This is not far fetched, say that the PSTN transitions to SIP and say that
peering on a bill and keep basis must extend to video conferencing in IP
and not only to voice, then once the bill and keep is IP, it makes sense
to run the rest of the IP traffic there, including the P2P traffic
transitioning through regional AS.
So my thought is that we can enable MH, but if we keep it regional and
then we do not have to worry about dampening lobotomized to handle
load of O(N*F)
F.
--
francois@xxxxxxxxxx
819 692 1383
On Thu, 30 Mar 2006, Christian Huitema wrote:
Dampening is part of the protocol and has nothing to do with the speed
of light.
Well, not really. Assume a simplistic model of the Internet with M
"core" routers (in the default free zone) and N "leaf" AS, i.e. networks
that have their own non-aggregated prefix. Now, assume that each of the
leaf AS has a "routing event" with a basic frequency, F. Without
dampening, each core router would see each of these events with that
same frequency, F. Each router would thus see O(N*F) events per second.
Since events imply some overhead in processing, message passing, etc,
one can assume that at any given point in time there is a limit to what
a router can swallow. In either N or F is too large, the router is
cooked. Hence dampening at a rate D, so that N*F/D remains lower than
the acceptable limit.
Bottom line, you can only increase the number of routes if you are ready
to dampen more aggressively. There is an obvious "tragedy of the
commons" here: if more network want to "multi-home" and be declared in
the core, then more aggressive dampening will be required, and each of
the "multi-homed" networks will suffer from less precise routing, longer
time to correct outages, etc.
There are different elements at play that also limit the number of core
routers. Basically, an event in a core router affects all the path that
go through it, which depending on the structure of the graph is
somewhere between O(M*log(M)) or O(M.log(M)). In short, the routing load
grows much faster than linearly with the number of core routers.
-- Christian Huitema
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf