North American Network Operators Group

Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical

Re: Deaggregation Disease

  • From: Joe Abley
  • Date: Fri Jul 21 10:49:36 2006

On 21-Jul-2006, at 09:17, Rob Evans wrote:

There seem to be a whole load of ASNs that have deaggregated.  AS5416,
AS5639, AS6140, AS9121, AS13049, AS16130, AS17849,  AS18049 (that's as
far as I got before getting bored).  Some of these are advertising the
covering prefix too, so they're certainly aware of how to aggregate.

Sometimes this is done intentionally -- the long-prefix (covered) prefixes might be TE routes designed to draw traffic to particular sinks through specific external providers.

People have been known to stamp NO_EXPORT on those and get some measure of TE without polluting the global table, but if the AS whose exit you're trying to influence isn't adjacent that doesn't work.

As it happens, Tony Li, Rex Fernando and I wrote up a proposal for a new attribute which might help in some of these situations. (It's a crude mechanism, but not as crude as NO_EXPORT). pathlimit-02.txt

Under that proposal you could stamp an AS_PATHLIMIT=2 or AS_PATHLIMIT=3 on TE routes and have them automatically dropped by routers when the AS_PATH length exceeded 2 or 3. For some people this would work (for others, for whom 90% of the Internet is less than 2 or 3 hops away it wouldn't do much).

It would help immensely with getting that document published if people could read that draft, and let me know if it looks like something they would implement if it was implemented. Private mail would be great.