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

Re: Route analysis from today's AS9121 incident - preso in Seattle?

  • From: Richard Irving
  • Date: Fri Dec 24 11:58:11 2004 
ren wrote:

Dear NANOG Program Committee,

Request: May we please have a presentation by Renesys on today's AS9121 incident at the Seattle NANOG 15-17 May 2005?

<soapbox>

   So... remember the previously posted comment that
most router melt downs are from human error ?

  Prophetic, isn't it ?

 It is sort of like when the Newbie unix Engineer comes up
and says "I can't get -so and so- application to work right"...

 And your aged -nix Guru, without even looking up, says "It is probably
a permissions error"..... and he is right.

  The true artform of the craft, is to realize that the maintenance,
and day to day operations of complex computer systems -isn't- a battle against
logic and math, like we were originally taught....

  But, in sooth, is a battle against human error,
in all its myriad of forms.


</soapbox>
Reason: This morning, 24-Dec-04, I gather AS9121 originations were spread throughout the prefix space. We, the peering community, noticed substantial instability throughout our peers as more specifics were accepted via peers without filtering. Given holiday staffing it took several hours longer to restore from max prefix states than it should have, and this is a problem hundreds of NOCs dealt with not just a handful.

It would be of use to discuss what happened, the prevention methods available and how to help feed the data collection for real time analysis. The Las Vegas NANOG is perhaps too soon so I am asking that this presentation be considered for the NANOG to follow in Seattle.

I'm requesting this in the open with the understanding past requests for Renesys data presentations have been turned down by the program committee. The 'can't prefix filter peers' problem has to be solved. NANOG is the right forum for this topic.

Sincerely, -ren