On Wed, Mar 5, 2014 at 1:20 PM, Joel M. Halpern <j...@joelhalpern.com> wrote: > 1) The set of traffic allowed is more nuanced and varied than just the two > categories of transit and peer.
Hi Joel, Barring a specific example, I respectfully disagree. Service providers often use BGP communities to give transit customers nuanced control over their routing, however those are still simple transit business relationships. If you pay someone for access to networks who aren't paying him, it's a transit link, however nuanced your actual use may be. Likewise, peering can be partial (typically regional) or full but either way it functions as a peering relationship, nothing else. > 2) If I have understand your description properly, it looks similar to the > frequent "valley free" assumption. And there are multiple papers that make > it clear that the actual paths are not "valley free". (Sorry, I do not have > pointers to the papers. I am not actually a researcher.) No problem, Google is an old friend. My understanding is that such valleys are a result of misconfiguration, not an intentional part of the system. Nor does the desired function of the system depend on their presence. Do you disagree? Regards, Bill Herrin -- William D. Herrin ................ her...@dirtside.com b...@herrin.us 3005 Crane Dr. ...................... Web: <http://bill.herrin.us/> Falls Church, VA 22042-3004 _______________________________________________ rrg mailing list rrg@irtf.org http://www.irtf.org/mailman/listinfo/rrg