Routing Statistics

Tue Jun 11 18:17:03 UTC 1996

At 09:49 AM 6/11/96 -0400, you wrote:

>	I went to NANOG meeting in Wash. DC a couple of weeks ago .
>	It didn't specifically address these issues , but there
>	was a presentation with following in-depth :) class on
>	peering with RS at MAEs/NAPs.Also lots of interesting stuff
>	were delivered by Merit ppl ( such as BGP peering stats etc).
>	The whole thing in general was very useful.
>
>	Rashid

I am still somewhat curious as to why exactly withdrawls are sent out for
routes that aren't announced.  I know its according to spec, but /why/ is it
speced that way.  It seems to me that the spec states that you should
withdraw routes which you do not announce then we may want to base route
instability only on the number of announcements you send and /not/ on the
number of withdrawls (as you are not creating instability by announcing
withdrawls which are passed on to you, only redistributing it).  In my
somewhat less than humble opinion having a large number of withdrawls, but
not a large number of announcements only means that you are exceedingly well
connected and are hearing routes from just about everyone and are therefore
are passing on all of their routing instabilities.

Justin "I hope that was coherent" Newton

Justin Newton
Internet Architect
Erol's Internet Services