Such engineering recommendation coming from a research group would be rather difficult to defend against these communities. Indeed, BGP/path vector protocols limitations being known, these WGs are actually seeking for understanding their impact (GROW) and advise BGP improvements to mitigate them (IDR). Imho, the main challenge for these WGs is to assess whether a proposed improvement has or not hidden negative impact(s).
In this respect, the reply would be "WHY" such recommendation ? and one would fall short on the answer because 1) it is difficult to prove that no workaround could ever be found that would (at least partially) solve some of these limitations and 2) afaik, RRG never actively worked on such issues (do we have experimental results arguing in favor of this recommendation). Note: this thread is imho representative of an RRG looking at protocol engineering recommendations instead of looking at experimentation/exploration of alternatives, their associated challenges (such as loc.reach, caching, mapping, impact on upper layers/traffic, etc.) and documenting them. -d. > -----Original Message----- > From: [email protected] [mailto:[email protected]] On > Behalf Of Ross Callon > Sent: Tuesday, December 30, 2008 6:55 AM > To: Joel M. Halpern; William Herrin > Cc: RRG > Subject: Re: [rrg] Consensus call: Reject proposed strategy F? > > > > > > I say we should tell the folks in GROW that it's time to stop > > > seriously entertaining the idea that vanilla BGP (strategy F) > > > will work out ok. What do you say? > > > > Firstly, given that there are useful things to do with BGP, > > that will extend its life, I don't think it would at all be > > appropriate to tell GROW not to put energy into engineering > > improvements. They are focussed on what the short term > > problems are, and the easily engineerable fixes (often just > > in terms of BCPs, such as Paul's FIB tricks.) > > I agree with Joel that it would be seriously inappropriate to > tell GROW > (or IDR) to stop thinking about engineering improvements to BGP. > > I can still remember quite well a comment from a "Gateway > Algorithms and > Data Structures" meeting (the group that preceded the IETF) in the mid > 1980's to the effect that "obviously IP routers will not work when the > IP forwarding table gets to 10,000 entries". I have similarly been > hearing about the imminent failure of BGP for nearly 20 years. Thus I > was originally tempted to just ignore this email exchange as "another > prediction of doom". However, I think that it is important to > point out > that the predictions of failure are neither new nor universally > believed. > > The fact that the RRG is exploring other options seems useful > to me, and > if nothing else can increase our understanding of what other options > might be possible. However, I am certainly not ready to confuse > "exploring" with either confidence that we want to change the > architecture, or consensus on how or whether to do so. > > Ross > > _______________________________________________ > rrg mailing list > [email protected] > https://www.irtf.org/mailman/listinfo/rrg > _______________________________________________ rrg mailing list [email protected] https://www.irtf.org/mailman/listinfo/rrg
