North American Network Operators Group Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical Re: Route table growth and hardware limits...talk to the filter
On Sat, 8 Sep 2007, Andrew - Supernews wrote: > >>>>> "Forrest" == Forrest <[email protected]> writes: > > Forrest> Sure, this would fail if a network decided to only announce > Forrest> /24's for example without a larger aggregate, but how many > Forrest> networks are really doing that? > > More than you probably imagine. > > Consider the following table: > > asn | count | c24 | c23 > ----------+-------+------+----- > 9583 | 1100 | 1014 | 42 > 7018 | 1140 | 764 | 125 > 17488 | 560 | 557 | 0 > 19916 | 568 | 532 | 6 > 701 | 729 | 501 | 38 > 1221 | 474 | 380 | 14 > 1239 | 572 | 359 | 24 > 577 | 417 | 337 | 19 > 209 | 587 | 329 | 38 > 17557 | 291 | 270 | 1 > 10292 | 270 | 267 | 2 > 4802 | 345 | 259 | 25 > 6140 | 315 | 243 | 16 > 4323 | 483 | 242 | 12 > 7474 | 284 | 228 | 8 > 2386 | 295 | 218 | 9 > 3301 | 307 | 217 | 30 > 702 | 392 | 206 | 34 > 6746 | 279 | 200 | 41 > > In this, "count" is the number of prefixes originated by the AS that > are not covered by any longer prefix (without regard to origin); "c24" > is the number of those prefixes which are /24s; "c23" is the number > which are /23s. I've cut this table off at 200 - the total number of > uncovered /24 routes across all ASes is 51811. > > Some of the above numbers would be worse if not for the presence of > over-large route announcements from other providers (for example, > Chinanet announce 125.96.0.0/14 even though 125.99.0.0/16 belongs > to hathway.net in India (AS 17488); approximately 230 _more_ /24 > routes announced by Hathway are in this range). > > You're right, that's way more than I would have imagined. Ok, why not combine the idea of throwing away more specific routes that have the same AS path as the larger aggregate with a mechanism that will do something like the CIDR-REPORT and aggregate bunches of routes that all have the same AS path. Or is the processing power/memory just not available to accomplish that? It seems either option would be better for not breaking connectivity than to simply reject anything longer than a /21 in 64/7 for example. Forrest
|