North American Network Operators Group
Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical

Re: /24s run amuck

  • From: Daniel Senie
  • Date: Wed Jan 14 17:00:38 2004 
At 03:36 PM 1/14/2004, Daniel Golding wrote:



Sadly, the type of person that public shame would work on, is the type of
person that is already taking care of the problem, or will be soon.

There is one mechanism for helping to solve this. Is there an RFC,
informational or otherwise that clearly specifies that BGP announcements to
peers and transit providers must be aggregated to the greatest extent
possible? If not, someone should write one. If yes, they lets publicize it.
This is a wonderful tool for network engineers to take to their managers, so
they can say "look, we have to do this, the RFC says so, and we MUST be RFC
complaint or #insert-horrible-thing will happen to us".

We live in a world of PHBs (Point Haired Bosses - see dilbert)
When those engineers succeed with their bosses in keeping RFC 1918 addresses off the backbone (read it, 1918 is a BCP, and says that), or when those engineers manage to implement RFC 2827 -- ingress filtering (also BCP) then maybe you'll have some ammunition that having a BCP about prefix filtering will be respected.

RFCs make suggestions. BCPs make stronger suggestions than some other RFCs, but clearly much of the community doesn't care, and ignores them just the same.