Connect Working Group Minutes RIPE 80

13 May 2020, 13:00–13:45 / 14:00–14:45
WG co-Chairs: Remco van Mook, Will van Gulik, Florence Lavroff
Scribe: Marco Hogewoning, Ferenc Csorba
Status: Final

1. Opening

Connect WG Co-chair Remco van Mook welcomed attendees and ran through the agenda. 

The minutes from RIPE 79 were approved.

2. Peering (Dis)Agreements during Covid-19 Times

Will Van Gulik

Presentation available at:
https://ripe80.ripe.net/presentations/25-Peering-aggreements-ripe80_public.pdf

Remco asked if they were now basically waiting for the Swiss regulator to set a tariff on peering.

Will said exactly – at least for the peering toward Swisscom. This also covered the period up to 2016, which was interesting because now that Swisscom no longer had an agreement with DTAG, the situation was a bit different. So maybe this would affect the situation, and that’s why they were waiting for the lower-instance Tribunal. 

Remco, asked if this would impact all peering in Switzerland going forward. Because if there there was a regulated tariff, he could see this being applied to everyone and everything. He asked if there was any language in the verdict about that. 

Will said he hadn’t seen anything like that. Most people in Switzerland did eventually do free peering where they just connected to the IXP and peered. He thought if you wanted to put a fee on access to your network, this was the case where it would occur. But he didn’t see anyone willing to do that, besides Swisscom and possibly UPC. 

Marco d'Itri, Seeweb, noted that Swisscom had over 50% of the access market. He asked why the regulator did not affirm that this was a dominant position, even after 2016? 

Will said this was an interesting question and he had been wondering the same thing. He recommended that Marco read the legal document. This was a bit beyond his own knowledge in legal terms, but he hadn’t seen any mention of this. 

Luca Cicchelli asked how much paid peering was used in Switzerland. 

Will said he could only answer for himself, but he did pay for peering with one of his companies, so it was used and there was a good service for that, but only from Swisscom as far as he knew. He didn’t know if other big ISPs applied this. 

Fredy Künzler, Init7, noted that Swisscom and DTAG had formed an illegal cartel until Jan 2016, which was not highlighted enough. He mentioned that the most important thing in the court decision was that the court acknowledged that "an ISP had a technical monopoly over his user base". This was new.

Will agreed that this had been stated in the document, which was something really strong in terms of legal language.

Carsten Schiefner asked if this could set a precedent for other (European) countries – even if Switzerland was not an EU country.

Will said he didn’t know. It would depend on how the other countries thought about it. He thought they could have an interesting precedent for other countries to push the incumbent to provide decently-priced access to their network as a peer.

Arturo Servin asked if this was the first peering battle to go to court or if there had been others in the past.

Will said there were no other court cases in Switzerland as far as he knew.

Remco said he knew of a lot of disagreements between large organisations that had brushed against the courts. However, it was the first verdict he had seen that explicitly referred to a regulator to set a tariff. He suggested they get an update on this topic at the next RIPE Meeting, as it would be interesting one for them to follow.

Will agreed and said he would love it if any of the involved parties could share their perspectives as well. 

3. The Challenge of Operations under Covid-19 Restrictions

Benjamin Schilz and Raphael Maunier

Presentation available at:
https://ripe80.ripe.net/presentations/26-Volterra-Ripe-connect-presentation.pdf

Julien Escario asked if they provided all ties, screws and “little stuff”?

Benjamin and Raphael replied that they did.

Ondřej Caletka, asked where their lab was located (given the company was designed as “remote-first”).

Benjamin said saying they had two racks in Equinix Paris with the lab and their equipment, and had stock in their office in the Paris suburbs.

Vasilis, Magma, asked how they ensured there no backdoors, accounts with admin capabilities or anything fishy introduced during the process.

Benjamin said they were shipping equipment with basic configuration and once they were able to connect to the network they erased everything.

Aleksei Sohanen, Trex, asked if they were planning to redo the site with new hardware after restrictions were lifted.

Raphael said this had been a basic deployment and they were planning further changes in September with more servers and an MX10k.

Andy Davidson asked if they would keep their detailed installation manual up to date. He also asked if they would do remote installations by choice in the future.

Raphael said from a business point of view, the cost to send people and have staff not working for a week was greater than the cost of doing a remote installation. They were negotiating with Equinix a “package of installation” that would be a project of deployment with a set cost, and as soon as they needed another PoP they would ship it.

Theo Vos asked what they had done with the flight cases once they were finished.

Benjamin said they had them shipped back to Paris.

End of first session.

5. What's going on? Lies, Damned Lies and Public Traffic Statistics

Remco van Mook and others

Presentation available at:
https://ripe80.ripe.net/archives/video/336

James Blessing, Jisc Collections and JANET, asked if the problem was a lack of information or a lack of knowledge on the part of journalists?

Remco said he thought it was both. They couldn’t blame journalists for going to sources of information that they knew were there. He also didn’t think it was fair for them to say that journalists shouldn’t be doing this, but if they wanted to stop them painting wildly inaccurate pictures then they should give them something better. 

Blake Willis remarked that journalists were used to contacting companies and asking them for comment.

Remco said this was true, but this didn't give the whole picture in aggregate. You could contact your local cable provider and ask how their Internet was working, but you would only get an answer from their marketing department saying “We’re fine, thanks for asking.”

Eileen Gallagher mentioned that journalists would sometimes use a graph to make judgments that shoot the narrative.

Remco agreed. There were graphs from Internet exchanges that could prove anything – a measurement could get missed and traffic would jump off a cliff, for example.

Nurani Nimpuno said it depended on how you defined “transparency”. The question could be how much traffic is sent and where vs whether it works and how did they know it would continue to work (especially in a crisis).

Remco van Mook said the second one “does it work and how do we know” was a fair one. Until now, what had been used as a proxy for that answer was traffic statistics. He thought they might be too secretive about sharing information that would help them answer Nurani’s second category of questions as an industry.

Daniel Karrenberg, RIPE NCC, said this was not a new problem – there had been many scares about the Internet in the past. The solution was to have a neutral point that could aggregate and supply the data. As self-serving as it might sound, he thought that the RIPE NCC filled this role.

Micheal said many journalists struggled with a lack of capacity in terms of both Internet knowledge and funding. But there were plenty of journalists and media organisations in the sector that the RIPE NCC could work with. For example, he was part of a new coalition within the IGF that was focused on journalist sustainability. Many of the media involved there could be a good mediator between the technology side and the media.

Remco said this could be very helpful once they had figured out what the technical side looked like.

Blake suggested that if anyone was approached by a journalist, the primary challenge had been in terms of physical vandalism of infrastructure and threats to technicians.

Remco said he liked the irony of people complaining that Facebook wasn’t loading their 5G disinformation fast enough while they were also burning down cell towers. This was general idiocy and wasn’t unique to their industry, it was a broader issue that needed to be addressed on a global scale.

Aaron said that Open-IX was releasing an interconnection report with measurement data that would be consumable by wide audience without heavy technical knowledge.

Remco asked him to share that on the WG mailing list once it was published.

Anita Nikolich, ARIN, shared a link to an opinion piece by Vice Cerf on this topic: https://thehill.com/opinion/technology/497053-the-internet-isnt-broken-but-its-inequalities-need-to-be-fixed

Remco said it would be good to share this on the mailing list as well.

Andy said in some markets the media's understanding of the Internet was shaped by customer complaints about lack of availability in some areas and unreliable or low performance. He thought they could demonstrate in the interconnection world that their services helped ISPs and content networks deliver unusual or unpredictable traffic patterns and that the debate needed to move to the last mile.

Remco agreed and thought they could help that debate by showing that they were really doing their jobs on their side. 

6. Impact of Covid-19 on Network Performance

Massimo Candela, NTT

Presentation available at:
https://ripe80.ripe.net/presentations/32-ripe80_covid.pdf

Vesna Manojlovic, RIPE NCC, pointed to some studies posted on the RIPE Labs:

Thomas King, DE-CIX, asked if they had any measurements from before the COVID‑19 lockdown, as the difference would be interesting to see.

Massimo said the entire thing was based on this. From the first week (in February 2020) Europe was still largely "normal" and the graphs used this as their baseline. The dotted lines in his graphs indicated when various lockdowns were enforced.

Rinse Kloek said they saw an enormous increase in Netflix traffic during lockdown. Once Netflix changed some of their video profiles, there was 50% decrease in Netflix peak. Another thing they saw was the 9-10am upstream peak from home.

Remco noted that the European Commission had asked Netflix and others to reduce their traffic profiles. He asked Massimo if this was reflected in his findings.

Massimo said that they didn’t have Netflix-specific measurements but their findings seemed to match these observations.

Remco noted comments from Swedish participants who clarified that while their country wasn’t on an enforced lockdown, it was on a recommended lockdown and many people had respected that.

7. Covid-19 IXP Update

Bijal Sanghani, Euro-IX

Presentation available at: 
https://ripe80.ripe.net/presentations/27-ripe80-covid-ixp-1.pdf

There were no questions

8. Covid-19 IXP Update

Nurani Nimpuno, LINX

Presentation available at:
https://ripe80.ripe.net/presentations/24-RIPE80-Peering-toolbox-Nimpuno.pdf

There were no questions.

End of second session.

RIPE Forum

The RIPE Forum is an additional way to participate in RIPE community mailing list discussions using a web-based interface rather than an email client.

Check out the forum