You're viewing an archived page. It is no longer being updated.
Connect Working Group - RIPE 72
25 May 2016, 11:00-12:30
WG co-Chairs: Remco van Mook, Florence Lavroff
Scribe: Emile Aben
The session started with Remco welcoming the room.
Remco proposed the chair selection process, which is the same process as the Cooperation Working Group is using, and called for consensus on this procedure.
Remco declared consensus on the chair selection process and asked the scribe to carefully minute that.
Arnaud Fenioux, France-IX
The presentation is available at:
https://ripe72.ripe.net/presentations/97-RIPE72_05-16.pdf
There were no questions.
Daniel Kopp, DE-CIX
The presentation is available at:
https://ripe72.ripe.net/presentations/92-2016-05-25_RPKI_Origin_Validation_at_IXPs.pdf
Alex Band (RIPE NCC) asked what toolset they are using for RPKI validation, and if it is working for them.
Daniel answered that they were using the NCC toolset, and it's working well, but he's not using it in a production environment.
Alex elaborated that he'd like the tool to be better.
Jen Linkova (Google) asked if he had IPv6 data as well.
Daniel said the analysis was only about IPv4, for lack of time he didn't do it for IPv6 as well.
Jen also asked if a lot of the wrong origin AS cases were private AS numbers?
Daniel answered he didn't look into that.
Tim Bruijnzeels (RIPE NCC) asked how bad it is to have invalid prefixes, if they have a covering valid prefix? He elaborated that there are instances where there are invalids made on purpose, so they are not necessarily a data quality signal.
Daniel agreed with this consideration.
Malcolm Hutty (LINX) said what really matters is what percentage of invalids are actually hijacks as opposed to something else.
Ruediger Volk (Deutsche Telekom) said invalids that are observed are useful to who publishes them, and that the benchmark of what is correct is the RPKI itself.
Ruediger asked if they have considered telling the IXP members community about the invalids seen through them.
Daniel thought that was a good idea.
Ruediger thought DE-CIX could use the ARIN trust anchor for this, as the relying party agreement allows for publishing of statistics if they are not easily machine readable.
Remco did a poll on getting more space for connect-wg at the next RIPE meeting. The room supported this.
Remco took a picture to support his claim to get more space for the working group.
Greg Hankins, Peering DB Board
The presentation is available at:
https://ripe72.ripe.net/presentations/81-ripe-72-peeringdb.pdf
Joe Provo (Google) asked what the default visibility for the imported data was.
Greg said he thought it was open, for those who don't want that, please correct it.
Sebastian (NSZR) asked if a search facility based on IP address space has been considered, which would be especially useful to researchers.
Greg said that would be a good feature.
The presentation is available at:
https://ripe72.ripe.net/presentations/117-Urricariet-Latest-Trends-in-Optics_RIPE-72_RevC.pdf
Paulo Lucente asked why he was exporting statistics with sFlow, and not IPFIX?
Christian said he'd have to ask others.
Gaurab Raj Upadhaya (Limelight) asked the audience if anybody was using 100G CLR4, because he wanted to exchange experiences. Nobody in the room reacted to that.
Tom Hill (Bytemark Hosting) asked when SPF 56 would be available.
Christian answered that deployment to market would be around 2018.
The presentation is available at:
https://ripe72.ripe.net/presentations/149-20160525-RIPE-AN-euroix-update.pdf
There were no questions.
Barry O'Donovan - INEX
The presentation is available at:
https://ripe72.ripe.net/presentations/110-Asymmetric-Routing.pdf
There were no questions.
Martin Levy - CloudFlare
The presentation is available at:
https://ripe72.ripe.net/presentations/124-RIPE72_Copenhagen_CloudFlare_Martin_Levy_RS_count.pdf
There were no questions.
At the close of the session Remco asked for feedback on the content and organisation of the connect-wg.
There was no feedback.
Remco closed the session