On Sun, Jan 4, 2009 at 12:48 AM, Tony Li <[email protected]> wrote: > |The history of the computing industry is one of regularly encountered > |one-time advances. Moore's Law isn't built on the back of purely > |evolutionary improvement. > > True, but one time advances from the networking industry catching up to the > rest of computing are all expended. It now tracks it quite well, albeit
So routing improvement now "tracks quite well" with general computing improvement which is predicted by Moore's Law to compound at 40% annually, just like I claimed for routers? > |2. Engineer's maxim: if it ain't broke, don't fix it. If failure does > |not loom on BGP's horizon, there is little justification for moving > |any disruptive strategy forward from research to engineering. The > |desirable output from our efforts is *very* different depending on > |whether or not BGP is failing. > > Again, it's not BGP, it's the number of prefixes. Any other protocol would > have to deal with the implications of a DFZ with a full set of prefixes as > well. Thus, it's the architecture that must change. Please take it as a given that "BGP failing" is in fact shorthand referring to the failure of the routing architecture in which every multihomed system requires a slot in the routing tables of every core router with all the attendant data transmission and computation. Et cetera. > And, once again, if > you don't believe that it needs to be fixed, I refer you to the problem > statement. I've long been convinced that BGP's scaling problem -should- be addressed. Whether it -needs- to be addressed is an entirely different issue which has a major practical impact on the intermediate research goals and on the appropriate research posture that focuses our efforts. Stance matters. Proposals based on the presumption that BGP -needs- to be replaced will far overreach proposals based on the proposition that we should steadily nudge the architecture towards something more desirable. Without a clear case that BGP -needs- to be replaced, that overreach will almost certainly render the proposals into operational non-starters. As we overreach, we'll tend not to closely examine the more moderate proposals for evolving in to an improved architecture that do not immediately form into a comprehensive solution. Tony, you've long complained that we're jumping straight to concrete action proposals instead of doing research, but the research stance based on the proposition that BGP will fail compels us to focus on comprehensive solutions, for which concrete action proposals are required. > |However frightful the monsters under the bed may be, there is > |insufficient proof of their existence to justify giving your toddler a > |shotgun. > > Hopefully there are better solutions than that regardless of motivation. My > toddler has horrible aim. ;-) When we asked the RIRs obstruct PI prefixes for multihomed IPv4 and IPv6 users, we upgraded them from cap guns to bullets. It's a very *harmful* practice which should be discouraged absent testable and quantitatively substantiated growth predictions projecting routing table growth which outstrips our ability to build hardware. The IPv6 side is especially harmful, given the looming free pool crisis whose projections are very well grounded. As a group, we should withdraw that recommendation and discourage obstructing PI space for multihomers. Regards, Bill Herrin -- William D. Herrin ................ [email protected] [email protected] 3005 Crane Dr. ...................... Web: <http://bill.herrin.us/> Falls Church, VA 22042-3004 _______________________________________________ rrg mailing list [email protected] https://www.irtf.org/mailman/listinfo/rrg
