IPv6 Working Group Minutes RIPE 73

IPv6 Working Group - First Session

Date: 26 October 2016, 15:00 - 16:30
WG chairs: Jen Linkova, Benedikt Stockebrand, Anna Wilson
Scribe: Alun Davies
Status: Final

A. Administrative Matters [5 min]

Welcome, thanking the scribe, approving minutes, etc.

B. IPv6 and the DNS [30 min] - Geoff Huston, APNIC

This presentation is available at:
https://ripe73.ripe.net/presentations/37-2016-10-24-ripe-ipv6wg.pdf

Geoff pointed out that he expected someone was going to ask about setting UDP and TCP to different values.

Anand Buddhdev (RIPE NCC) made a comment about ‘glueless delegations'. He pointed out that glue is not entirely optional in some cases, although it was optional in the context of Geoff's experiment. Geoff agreed, but added that there are a small number of resolvers that don't work well with glueless. He added that, as long as route zone operators pay careful attention to the experimental setup in relation to these issues, the results should be pretty good.

Alain Durand (ICANN) commented that RFC 3901 was written in September 2004 on DNS IPv6 transport guidelines. He asked whether the document is still current, or needs to be revisited.

Geoff replied that if the doc didn't talk about systems like UDP and TCP flow control, it should. He added that they needed to do more v6 quick work to get clear on the full extent to which such a document might need to be revised.

C. Measuring the Effects of Happy Eyeballs [15 mins] - Vaibhav Bajpai, Jacobs University, Bremen

This presentation is available at:
https://ripe73.ripe.net/presentations/125-index.pdf

Geoff Huston asked how network operators know when to turn off v4 completely, and how to know if the v4 traffic is redundant if you've got v4 and v6 on dual-stack. He added that the point of happy eyeballs (HE) is that both protocols can be used in dual-stack but that in normal circumstances, v6 gets chosen. HE was not designed to pick the fastest - this was only a happy side effect.

Vaibhav said that HE does not give preference to v6.

Geoff answered that as long as v6 is working using HE, v6 is preferred.

Jen Linkova, WG Chair, commented that she agreed with Geoff that with HE, it's not a matter of speed but a matter of avoiding timeout.

Vaibhav pointed out that in his paper, in agreement with both Geoff and Jen, that we still need HE.

D. IPv6 Deployment in Countries in Relation to their GDP per Capita [30 mins] -Alain Durand, ICANN

This presentation is available at:
https://ripe73.ripe.net/presentations/101-IPv6-GDP-ripe73.pdf

Jordi Palet Martinez (Consulintel) commented about countries with slower GDP having more incumbent operators.

Alain agreed that governments have an impact on the issue.

Marco Hogewoning (RIPE NCC) commented that the study was very interesting and worthwhile.

He said he'd done a similar one a few years back for a talk for the OECD. Marco said it's interesting to ask whether richer countries do better with v6 deployment, and for future studies, it might be worth considering full Internet penetration of the countries and not just IPv6. He asked if more Internet in the country increases chances of IPv6 and suggested that GDP is not a very big factor in whether you do IPv6 or not. It's open to the group to figure out what the big factors are, but in his view it's probably not the money.

Benedikt Stockebrand, WG Co-Chair, said one factor in deploying IPv6 is money, but it also takes some provider or whoever to run into problems (he referred to an issue Comcast had in not being able to run their boxes anymore). He added that you needed a persistent voice advocating IPv6 all the time, which can happen at a country scale. GDP is a factor, but you have to look at such factors together with others.

Alain Durand responded that it's always important to pursue further study and go back and look at outliers (e.g. countries where IPv6 works well despite expectations). This would make for a good grad level project.

Rob Lister (LONAP Ltd.) asked (via chat) whether Alain had looked at any other metrics besides GDP.

Alain replied that GDP was the first metric he tried. After talking to some economists, he changed to GDP per capita.

Rob Lister commented that newer countries with new infrastructures might actually be able to deploy IPv6 more easily.

Jen Linkova (WG Chair) added that there is probably some correlation between how rich a country was and how operators worked in the country.

Alain replied that it was a good point and that in many countries service providers are offering the cheapest thing possible. In rich countries, you may have relatively short cycles for equipment.

Jen added that at the last IPv6 conference, people were warned not to send used equipment to poorer countries.

Marco Hogewoning pointed out that the current Internet governance best practice forum for IPv6 is looking into other factors. He invited Alain to contribute to this. In response to Jen, he added that there is probably a correlation between the way providers operate in different countries and IPv6 deployment. It is probably worth looking at the market situation in countries that have unexpected IPv6 deployment.

Alain pointed out that such factors could indeed create peculiar effects and needed to be accounted for. There was some further discussion on the topic between Marco and Alain.

Mirko Mikovic (Serbian National Internet Domain Registry) raised questions about poor countries, particularly Serbia and Bosnia. He suggested that Bosnia has high v6 deployment because someone was brought in who knew how to do it. In Serbia, they are sticking to IPv4 because it's too expensive to migrate to v6.

Alain thanked him for his comments.

Blake Willis (L33 Networks) pointed out that it could be that global transfers are constrained to transfers between particular countries in particular regions. He pointed out that it would be hard to verify this through research, although perhaps easier if we look at v4.

Alain replied that, according to brokers he'd been in contact with, they are not only making transfers between particular countries. So transfers seem to be global now. The question to ask, he added, is how predominant legacy addresses are in particular countries.

Marco, in response to Blake, replied that he didn't have any hard data on this issue, and observed that there doesn't seem to be data to support Blake's suggestion.

Rob Lister (via chat again) pointed out that many countries in Africa are jumping straight to mobile and this may have some effects on v6 deployment.

Alain commented that these countries were not making a big spike in the graph.

E. Working Group Chair Selection [10 mins]

Jen Linkova informed the audience that Anna Wilson would be stepping down as IPv6 Working Group Chair. She invited the audience to thank Anna for all her hard work.

Benedikt Stockebrand added that it was Anna who really got him and Jen up to speed when they became chairs.

Anna gave a short speech thanking everyone, adding that she looked forward to seeing everyone at future RIPE meetings.

Jen then announced that Raymond Jetten (Elisa OYJ) would be taking up Anna's position as Co-Chair for the IPv6 Working Group.

Raymond made a short speech thanking everyone and pointing out that he had plans to make some changes to current attitudes on certain issues in the working group.

---------

IPv6 WG RIPE 73 (second session)

Date: 27 October 2016, 17:00 - 18:30
WG Co-Chairs: Jen Linkova, Benedikt Stockebrand, Raymond Jetten
Scribe: Mirjam Kühne
Status: Draft

A. IPv6 and IPv6-only Deployment in Microsoft - Marcus Keane, Microsoft

This presentation is available at:
https://ripe73.ripe.net/presentations/162-Microsoft-IT-IPv6-RIPE-Madrid-October-2016.pdf

Nathalie Trenaman (RIPE NCC) asked what was the part of the IPv6 deployment they actually had to spend money on.

Marcus responded that this was the NAT64 equipment, but that it was definitely worth it.

Wilhelm Boeddinghaus (IUBARI GMBH) mentioned that tunneling IPv6 on mobile phones has been quite successful. He asked if Microsoft has plans to implement software for office PCs with which they can then implement IPv6 only on the network and tunnel-to-tunnel IPv6 through it. This would ease the burden of deploying IPv6 in office LANs.

Marcus asked if he is referring to CLAT or to tunneling.

Wilhelm clarified that it didn't really matter. It should be a simulation on the IPv4 so they don't have dual-stack on the LAN segments. He added that there was a lot of old equipment that will never be running IPv6. And having some tunneling mechanism implemented and tested by Microsoft would be helpful.

Marcus responded that there were no such plans at the moment, at least not in the US.

Jordi Palet (Consulinte) added that the CLAT client it was not a tunneling mechanism, it was for translation. The right place to deploy it would be in the CPE. There are RFCs describing this. The idea is not to deploy it in every PC.

Herve Clement (Orange) asked about the interaction with the development teams at Microsoft and how possible issues are tackled. He added that he was interested to find out if Marcus was planning to develop any educational material for customers based on the experience gathered in the process.

Marcus replied that they have active conversations with the product groups. Regarding the second question, Marcus responded that they don't have any such plans. But he thought it was a good idea.

B. E-mail Services on IPv6-only - Ondřej Caletka, CESNET

This presentation is available at:
https://ripe73.ripe.net/presentations/139-e-mail_services_on_IPv6_only.pdf

Jan Zorz (Internet Society) commented that he thought this is was a brilliant service and he thanked Ondrej for doing this. He added that he's just been testing it. He watched the log files and noticed that there was some complex stuff going on in the background. He also noticed that not even the gray-listing confused the system.

Radu-Adrian Feurdean (MyStream/Coriolis) asked if Ondrej could gather data to present at a future RIPE Meeting about how different anti-spam systems react when sending IPv6 only.

Ondrej responded that he is not sure if he can automatically produce this data. He discovered this manually by looking into log files. It would take him a lot of time, but he said he could try to do something.

C. IPv6 Status at RCTS - Carlos Friacas, Fundacao para e Ciencia e a Tecnologia, I.P.

This presentation is available at:
https://ripe73.ripe.net/presentations/128-IPv6_at_RCTS_cfriacas_20161027.pdf

No questions

D. IPv6 Prefix Delegation for End-Users BCOP - Jan Žorž, Internet Society

This presentation is available at:
https://ripe73.ripe.net/presentations/106-Jan_Zorz-IPv6-prefix-delegations-BCOP.pdf

Nathalie thanked Jan for picking this task up and said that the RIPE NCC was asked this question a lot during the IPv6 training courses. She added that also in Jordi Palet's survey they see that people get the weirdest prefix sizes. She urged Jan and the community to get this BCOP document out ASAP.

It was agreed that people can and should send feedback to Jan directly. Jan will then incorporate the changes and send the new version of the document to the IPv6 WG list. Once the WG approves the document, it can be published as a RIPE document.

E. Open Mic: What Does Success Look Like? - Anna Wilson

This presentation is available at:
https://ripe73.ripe.net/presentations/181-ripe73-what-do-we-want.pdf

Wilhelm Boeddinghaus commented that he didn't think this was a sports game. Success is not about getting new supporters. Success is to get IPv6 in the networks smoothly without economic disasters. It was about economic success, not technical success. He added that the real goal is to have a smooth transition.

Oskar Stenman (Magine TV) said he thought they would be done when they don't have to care about IPv4 anymore.

Anna asked what the steps should be to get there.

Oskar said that it was a difficult question at this point.

Carlos Friacas commented that the goal was to achieve everything Anna listed in her introduction slides - and much more than that.

Anna said that she disagreed with Carlos. She added that focus was very important, otherwise, it wouldn't get done. The Alexa websites are probably not such a big deal compared to other things that would make a difference.

Benedikt Stockebrand said that there was no single goal. He wondered if perhaps the goal for the RIPE community would be to shut down the IPv6 WG. He added that success would be if anyone involved in networking assumes IPv6 when he or she hears IP. Everything else is nice, but not that important.

Sander Steffann said that agreed with Benedikt and added that success is the moment when someone cannot ignore IPv6 anymore. There are still products that come out without IPv6 support. The moment when it is no longer viable to sell products without IPv6 support should be considered a success.

Anna concluded that she heard two contradicting things: Wilhelm said that a smooth transition is important. Sander said that success is when IPv4 will be uneconomic.

Gert Doering (SpaceNet AG) responded that the goal is to make IPv4 have no value again. The fact that integers cost money today is wrong. The question is when this will be the case. He added that Geoff Huston had said it would be another four years before that happens but didn't think it would be smooth.

Jan said that there were three parts to IPv6 deployment: IPv6 awareness, IPv6 deployment and learning, and finally business as usual. The goal is to try to bring the rest of the world that is still in the awareness phase into the deployment phase. The awareness phase has been long enough. He added that they needed to convince people to deploy and they needed to document experience and best practices. For operators, the goal and success would be that users don't even notice that they are using IPv6.

Anna said she would like to explore who were the most effective people. Who moved already and what made them move.

Lasse Jarlskov (TDC) said he would like to go back and ask what is the goal of the DNS WG was since DNS was deployed. There wasn't just one goal. He added that there were short-term and long-term goals and steps need to be taken. This WG is taking steps into the right direction.

Herve added that he agreed with Gert and that one important metric is the size of the IPv4 routing table. When it start going down, then they're having success.

Radu said that he didn't agree with Jan that they should go out of the awareness phase, at least not until the time when IP means IPv6.

Anna thanked everyone for the discussion and noted that everyone has different goals and it would be difficult to converge them. She added that she would try to use the input and see how to move forward.