- content to the Chair of the working group.
- format to webmaster _at_ ripe _dot_ net.
7 May, 2009, 14:00 - 15:30
Chair: Joao Damas
Minutes: Franz Schwarzinger (RIPE NCC)
Jabber Scribe: Arne Kiessling (RIPE NCC)
A. Administrative Matters
The minutes from the RIPE 57 Meeting were approved.
B. The Intra-domain BGP Scaling Problem
During the presentation, when Danny mentioned that an average of 34% percent of the received updates are duplicates, Greg Shepherd asked if he was talking about duplicate paths or duplicate prefixes.
Danny McPherson replied that he was talking about an exact copy of every update.
After the presentation Greg raised concerns about adding attributes to iBGP which might affect scalability and functionality of the Internet at large. He asked why this is not yet being discussed at the IETF.
Danny agreed that this dialogue needs to take place and that scalability issues need to be taken seriously.
Volodymyr Yakovenko (Google) asked about implications of this regarding route reflection.
Danny answered that there are improvements being worked on by CISCO and that 95 percent of the effects can be eliminated.
Joao Damas (Chair) thanked the speaker.
C. Best-effort Internet Video
Lorenzo Colitti (Google) asked Greg if he'd thought of implementing an unreliable datagram transmission system on top of HTTP to get through firewalls, using the TCP sequence number as a packet sequence number and ignoring TCP ACKs. The discussion rapidly came to the conclusion that this was a bad idea to fix bad behaviour, it still wouldn't work with firewalls that perform deep packet inspection, and would require effort that could otherwise be put into fixing the broken middleboxes.
D. The C-BGP Routing Simulator
An attendee asked if this could also be used to model incoming traffic.
Bruno answered that this is a very common question. Most of their studies were focused on outgoing traffic. In order to get an idea on what is happening to incoming traffic, a model of the outside network would be needed. This presents a big challenge. One could use AS level topologies in order to approximate the outside network, but there will always be inaccuracies.
Volodymyr Yakovenko (Google) stated that this concept is very interesting. He asked if it is possible to understand internal topologies which are using traffic engineering.
Bruno answered that this is not possible since they are focusing only on IGBP.
Volodymyr also asked if it is possible to parse routing policies.
Bruno answered that it is possible to some extent, however complex policies might not work.
Volodymyr asked if they are taking different implementations and software version regarding routing policies into account.
Bruno answered that they only have a common implementation at the moment. They are not seeing a big need to support this. Limited resources also present a problem.
Y. I/O with Other Working Groups
Joao Damas (Chair) mentioned IPv6 policy changes proposed in the Address Policy Working Group that will affect global routing tables. He announced a session in the closing plenary where these issues will be addressed.
The session ended.