RIPE 61

EIX WG RIPE 61 Minutes:
RIPE Meeting: 61
Working Group: EIX
Status: Final
Revision Number: 1


RIPE 61 EIX Working Group – Session I and II
Rome, 18 November 2010, 09:00-10:30 and 11:00-12:30
Co-chairs: Andy Davidson, Fearghas McKay.

A. Agenda Bashing

  • Welcome

Fearghas welcomed the attendees and began the EIX session at 09:05 local time. He introduced a new format for the meeting, and announced that the usual IXP updates are now no longer a part of the EIX WG. He expressed a desire to include a summary document that includes a 1-2 page submission from IXPs and asked the audience to submit for the next meeting.

Fearghas also announced that Cara has retired from Co-Chair and thanked her for her work and wished her luck on her MBA. He said they were not currently seeking a new Co-Chair at this point. He said there’d be more news over the next six months.

  • Scribe and Jabber Monitor

Andy thanked the technical team, scribe, and Jabber monitor from the RIPE NCC for their help during the session.

B. Approval of minutes from RIPE 60

Minutes were approved without comment.

C. Peering in Italy

Fearghas said that two presenters for this session had not yet arrived,so the session might be broken up into two (before and after the coffee break).

Antonio Soldati gave a presentation on Telecom Italia’s peering policy. It’s available here:
http://ripe61.ripe.net/presentations/238-IP_Peering_Policies_in_TI_-_Intervento_Soldati_61_RIPE_Meeting.pdf

Fearghas announced prizes for any good audience questions (and promised that no one would laugh if the questions were bad).


There were no questions for Antonio.

Maurizio Goretti gave a presentation on NaMeX (Rome). The presentation is available here:
http://ripe61.ripe.net/presentations/337-NaMeX-IXP_WG_RIPE61.pdf

Mathieu Paonessa from Jaguar Network asked about NaMeX’s future landing station project and if NAMEX will make people install IP routers so they can peer on NaMeX or any other IXP

Maurizio said that there were a lot of IXP end carriers remotely connected to other IXPs and that they arrive in the landing station. He added that he thinks being present in a landing station could help the possibility that a potential customer will connect to NaMeX’s exchange and stop the IP traffic in Rome. He added that Italy is one of the most important countries for the sea cable landing station, but that most of the carriers just back-hole to northern Europe and they are trying to stop this traffic.

Mike Blanche from Google asked about commercial model for connecting, (because bandwidth isn’t free).

Maurizio said that their members ask NaMeX to optimize the cost to connect to AMS-IX. They do this by sharing the costs to lower them (i.e. buying circuits). They have also asked other IXPs to connect with them, but AMS-IX is the only one that lets you have a multiple MAC addresses on one circuit.

Mike asked if the fee was flat per meg.

Maurizio said it depends on the bandwidth they get. He said the relationship is not just political, it’s also about collaboration. The collaboration with a bigger IXP helps NaMeX to grow. He added that this wasn’t an exclusive agreement, it was something that came from discussion with the members and the board agreed.

Mike said it was interesting and that there were many attempts to build regional IXPs and this was quite unusual.

The session’s chat monitor, Fergal Cunningham from the RIPE NCC, asked a question on behalf of Alexander (no affiliation given): What’s the price to connect from NaMeX to AMS-IX?

Maurizio said it wasn’t appropriate to discuss this during the session but that he would be happy to discuss it offline.

Mauro Magrassi gave a presentation on MIX-IT (Milan). The presentation is available here:
http://ripe61.ripe.net/presentations/211-MIX_Milano_RIPE_LXI_EIX_WG_SEC_C.pdf

James Blessing from LimeLight Networks commented that MIX-IT and NaMeX were both building into landing stations in Italy and asked if it wouldn’t make more sense to work together rather than duplicating it all. He then asked if MIX-IT is actually building remote ports into the exchange in Milan or if they’re actually putting switching-infrastructure in those locations?

Mauro said they decided to put switches in that location because they feel it’s a bit early for that [switching-infrastructure], but the key point was that sooner or later, the fact that people coming from the different side of the Mediterranean basin should appreciate the fact that they can exchange traffic earlier than friends in Northern Europe.

James again asked about his earlier comment on the duplication of effort and Mauro said that they have many good restaurants to discuss such possibilities.

Luca Cicchelli gave a presentation on TOP-IX (Torino). The presentation is available here:
http://ripe61.ripe.net/presentations/114-TOP-IX_Oct10_RIPE.pdf

Mathieu Paonessa from Jaguar Network asked how the partnership would work if Lyon IX is partnering with both TOP-IX and IFX in Paris?

Luca said that at the moment, they are two different things. TOP-IX only has a staff of 12, so they prefer to work on a step-by-step approach. The only relationship currently established is with Lyon IX.

Fearghas said that the session was running over and that agenda item F., and maybe half of agenda item E would go into the next session.

D. Serge Radovcic gave the Euro-IX update.

The presentation is available here:
http://ripe61.ripe.net/presentations/338-Euro-IX61eix.pdf

There were no questions for Serge.

Fearghas then announced the next agenda item and said that they’d probably only get a couple done before the break, so he asked the presenters to be brief. Daniel Karrenberg from the RIPE NCC joked that they should offer t-shirts for the shortest presentations.

EIX Working Group II
Thursday, 18 November, 11:00 – 12:30

E. IXP News

LINX – “Architecture Refresh” – Kamran Khalid. The presentation is available here:
http://ripe61.ripe.net/presentations/112-RIPE61-LINX-architecture-review.pdf

There were no questions for Kamran.

NYIIX – Manhattan – “Experiences operating new Brocade platform and new location” – Akio Sugeno. The presentation is available here:
http://ripe61.ripe.net/presentations/212-NYIIX-RIPE61-Nov2010_Rev_2.pdf

There were no questions for Akio.

NIX.CZ – Prague – “Our altered topology” – Petr Jiran. The presentation is available here:
http://ripe61.ripe.net/presentations/202-NIXCZ-pj-ripe61.pdf

Nick Hilliard from INEX asked why they use a spanning tree when they have a virtualised core.

Petr said they don’t need the virtualised core, but if they want to connect some switch to two access switches you still one.

Nick asked if it was purely for edge loop detections and Petr confirmed that that was the case.

NIC.BR - “PTT Metro – Brazilian IXP (Newcomer)” – Milton Kaoru Kashiwakura. The presentation is available here:
http://ripe61.ripe.net/presentations/135-eix_wg2010.pdf

There were no questions for Milton.

SFINX – “Layer 2 interconnect, and new equipment” – Laurent Gydé. The presentation is available here:
http://ripe61.ripe.net/presentations/240-SFINX-update-RIPE61.pdf

Mathieu Paonessa from Jaguar Network commented that SFINX members were able to reach France IX members through the France IX VLAN, but it wasn’t possible to connect the opposite way. He asked if there were any plans to do this.

Laurent replied that it is planned that both VLANs run on both infrastructures. He added that members of both France IX and SFINX could already set up peering.

Mathieu then asked if SFINX was still planning to have some optical exchange point in Marseille (MORAN)?

Laurent replied that the project is still ongoing and he hopes he can share more information on this at RIPE 62.

Martin Levy from Hurricane Electric asked if they needed to sign a SFINX contract if they connect through France IX.

Laurent said no, the goal was to make it simple. You sign only the contract with the IXP you are connected to, and that they don’t charge
specifically for the people that use the interconnected. It’s provided as a new service.

UA-IX – Kiev – “The Secret of rapid traffic growth” – Sergi Polischuk. The presentation is available here:
http://ripe61.ripe.net/presentations/340-eix-ripe61-polishchuk-20101118.pdf

Mathieu Paonessa from Jaguar Networks asked if the local incumbent was peering with everybody or just small or new players.

Sergi replied that there is only one ISP in Ukraine, and they don’t connect to UA-IX because they won’t sign the UA-IX peering agreement.

Mike Blanche from Google asked if they would get more members/traffic if their rules were looser.

Sergi replied that he thought their rules gave them more traffic and that Google is also available at their exchange via UA-IX member, TopNet.

Volodymyr Yakovenko from Google asked whether UA-IX has done any research on what sort of traffic (like basic fingerprinting) people are sending through UA-IX’s exchange?

Sergi answered that they cannot look inside of our members' traffic, but it's the same as all Internet traffic.

Florian Hibler from EuroTransit commented that UA-IX’s website was only in Russian and asked if there were plans to translate it.

Sergi replied that the website content would be translated into English.

Fredy Kuenzler from Inet Seven AG commented that it was interesting that UA-IX forbid private interconnects and asked why this was the case.

Sergi said it was easy because most ISPs have looking glass, and that they step-in only when a member complains about a situation. He added that their main goal is to make the Internet connection better in Ukraine. If something goes wrong, then they take action.

F. Identify Yourself

Fearghas asked audience members wishing to participate in the 60-second to stand up. Six people stood up.

Job Witteman from AMS-IX talked about their planned 100 gig deployment.

James Blessing asked if the remote member signs the AMS-IX contract or if someone signs it in proxy?

Job answered that the partner signs in proxy; AMS-IX doesn’t charge the End User directly.

Nick Hilliard asked what AMS-IX was doing for the core, in particular, how they were integrating DWDM core with their 100 gig edge requirements.

Job answered that the 100 gig is customer connections only, it's not inter-switched connections. He added that they have eight switches in the core, and as load balancing they keep to multiple 10 gigs to get to the 8 cores.

Nick commented that that is not going to scale indefinitely and asked if there were long-plans to move to 40 for your individual bearers.

Henk Steenman from AMS-IX replied that they would start with 100 gig towards customers, and keep 10 gigs in the core until 2012 when it makes more economic sense to move the core also to 100 gig. He added that they didn’t expect larger than 10 gig streams that would offer any part of a LAC.

Frank Orlowski from DE-CIX gave an update. The presentation is available here:
http://ripe61.ripe.net/presentations/336-ripe61_decix_news.pdf

John Taylor from Equinix gave an update. The presentation is available here:
http://ripe61.ripe.net/presentations/259-Equinix_Update_RIPE61.pptx

Martin Hannigan from France-IX gave an update. The presentation is available here:
http://ripe61.ripe.net/presentations/263-France-IX-presentation-for-RIPE-meeting.pptx

Will Hargrave from LONAP Ltd gave an update. There was no presentation.

Nurani Nimpuno from Netnod gave an update. The presentation is available here:
http://ripe61.ripe.net/presentations/247-netnod-eix.pdf

Martin Fischer from ACOnet gave an update. There was no presentation.

G. Route Server RFC Update – Elisa Jasinska (Limelight Networks).

The presentation is available here:
http://ripe61.ripe.net/presentations/260-ripe61-draft-jasinska.pdf

Volodymyr Yakovenko from Google asked why she thinks that the removal of the system of route server AS number is a mandatory requirement.

Elisa said that it could alter the best path selection.

Volodymyr said he disagreed and asked to see slide 2 again and then realized he read the slide wrong.

Martin Levy from Hurricane Electric said he appreciated the work going onto the route server and said he never wanted to see the AS Number of a route server ever because its an anomaly of a policy, not of routing.

Daniel Karrenberg from the RIPE NCC said he was religious about being minimalistic about RFCs and asked for the motivation of the word ‘process’ in one of Elisa’s slide.

Elisa said that the motivation for the word "process" are cases where something is actually meant for the route server itself and it has to do something with it because that would be processing it.

Daniel asked why she wouldn’t want it if it doesn’t affect the output.

Elisa said it might affect the output.

Daniel said he understood.

H. Nick Hilliard (INEX) – “Small is Beautiful”.

The presentation is available here:
http://ripe61.ripe.net/presentations/339-inex-ripe-rome-eixwg-2010-11-18.pdf

There were no questions.

I – Open Discussion on the Switch Wishlist – Andy Davidson

Andy said that it was about time to go over the wish list of technical documents to be more in line with the needs of organisations that were buying their switches from Internet Exchanges today. He asked that the discussion be continued on the mailing list.

Mike Hughes said he was one of the original editors and said he completely supported changes or updates to the document. He commented he was happy at the evolution of the document and the power of it. Andy added that the conversation would continue at RIPE 62.

Fearghas urged the audience to subscribe to the mailing list, it’s how things get done. He asked that proposals be sent to the EIX WG Chair list, not to individuals. He also mentioned that RIPE 62 was now 2 – 6 May 2011.

Z. AOB