IPv6 Operations - IETF 101 Monday 9:30 Chairs: Fred Baker, Ron Bonica, Lee Howard Jabber: Mikael Abrahamsson Notes: Barbara Stark Agenda: Primary discussions Mythic Beasts: an IPv6-Preferred Data Center Requirements for IPv6 Routers 2018-03-04 , Using Conditional Router Advertisements for Enterprise Multihoming 2018-02-27, Brief discussions The following have been posted but had little or no comment on v6ops@ietf.org. We have a little time to talk, so we'll let folks talk briefly. IPv6 Performance Measurement with Alternate Marking Method 2018-02-26, IP Fragmentation Considered Fragile 2018-03-04, NAT64 Deployment Guidelines in Operator and Enterprise Networks 2018-03-05 , IPv6 Point-to-Point Links 2018-03-05, Transition Requirements for IPv6 Customer Edge Routers to support IPv4 as a Service 2018-03-02, IPv6 Prefix Delegation Models 2018-03-02, Draft Status: The status of v6ops drafts, both working group drafts (draft-ietf-v6ops-*) and individual submissions to the working group (draft--v6ops-*), may be determined from https://datatracker.ietf.org/wg/v6ops/. ----------------------------- Chair opening comments Fred Baker thanked Lee Howard for all he has done, since Lee is stepping down as a co-chair. Fred Templin will present first, because he is presenting remotely and it is 02:30 for him. Then proceed with agenda as posted. ---------------------------- Fred Templin presented IPv6 Prefix Delegation Models https://datatracker.ietf.org/meeting/101/materials/slides-101-v6ops-ipv6-prefix-delegation-models-00 Fred Baker: 10 minutes is up. Please comment against this draft on the list. ------------------------------ Pete Stevens presented IPv6 Only Hosting https://datatracker.ietf.org/meeting/101/materials/slides-101-v6ops-ipv6-only-hosting-00 Mikael: From jabber: question for the mic: how many customers insist they still have users only on IPv4, so they can't go to IPv6-only? Pete: v4 will never go away, but it's ok, because it's proxied. We don't care whether people are trying to reach the services over v4 or v6. Mikael: The question was how many of your customers are actually using v4? Pete: In the VM managed service, about half still have a v4 address. Lee Howard: Pete: The stuff that runs our network is proprietary. There are various challenges. Lorenzo Colitti: I think you're showing there really is a cost to running IPv4, and it's great to see that this is true now. Auwal Tata (remote): Fred: No one else at microphone. Thank you. ------------------------------- Fred Baker walked through Russ White's slides for Requirements for IPv6 Routers [starts around 1 hour into the meeting] https://datatracker.ietf.org/meeting/101/materials/slides-101-v6ops-requirements-for-ipv6-routers-00 Fred Baker: How do we move forward? Barbara Stark: The problem for me is in the scoping. Russ needs to state which set of routers he wants to address, and it can't be "all". I don't know what set Russ specifically wants to ensure is included in scope. I tried suggesting Enterprise, but he didn't like that. It didn't seem to correctly describe all the routers he was interested in. Lorenzo: One size does not fit all. And we should not have a shopping list. Consider conditional statements in requirements. Tim Chown: Agree with most of what Lorenzo said. Worked to harmonize this with 6434-bis. I like this draft and think it has a lot that is good. Fred: I will ask Russ to listen to recording and look at minutes. -------------------------------- Jen Linkova presented Using Conditional Router Advertisements for Enterprise Multihoming https://datatracker.ietf.org/meeting/101/materials/slides-101-v6ops-conditional-router-advertisements-for-enterprise-multihoming-00 Tim Chown: I think this is useful and would like to see it progress. I like the changes. Should 6434-bis make this a MUST instead of SHOULD? Jen: Leave as SHOULD. Fred: If there are cases where you wouldn't want to do this, then it's a SHOULD. Are there cases? Tim: Need to discuss in 6man. Lorenzo: Go for Last Call and get it done. I would like the doc to explain how this is similar to IPv4 with NAT. Jen: You're right. I can add some to section on limitations. Lorenzo: This is possible today if the network supports it. Because 5.5 doesn't exist in most hosts today -- meaning changes are needed to hosts anyway -- and it isn't the best approach / is a hack, should we figure out the right approach? Tim: I agree. Tatuya Jinmei: Ready for Last Call. Regarding 5.5, wondered if it may now be supported by more devices. Barbara: I support Last Call and like the draft. Juliusz Chroboczek: Is there a list of devices that support 5.5? Jen: Only know Windows supports. Fred: I will start a 2 week Last Call. Even if you have no comments, please indicate support. --------------------------------- Giuseppe Fioccola presented IPv6 Performance Measurement with Alternate Marking Method https://datatracker.ietf.org/meeting/101/materials/slides-101-v6ops-ipv6-performance-measurement-with-alternate-marking-method-00 Fred: When you say adopt by WG, is that v6ops, 6man, which WG? Giuseppe: I don't know. Fred 6man is this afternoon so decision to adopt work to define new bit will be there. Tim Chown: References are experimental. In 6man obvious concerns will be you can only set to random if arrives with zero value and many hosts are setting so your ability to re-mark may be limited. Lee: As a co-chair, I'd like to understand whether there is operational interest in using this? Fred Telecom Italia seems interested. Lee: Yes, but that's one. Michael Ackerman: As an enterprise operator it may be interesting. ---------------------------------- Ron Bonica presented on IP Fragmentation Considered Fragile https://datatracker.ietf.org/meeting/101/materials/slides-101-v6ops-ip-fragmentation-considered-fragile-00 Mikael: I think the network operator recommendation needs to be stronger. There are many networks where the ICMP message isn't even generated and sent. Ron: Right. I didn't think about that. That will be in the next version. Mikael: We need to include recommendation for support of PLPMTUD Wes George: The problem is technically we still can't do packets bigger than 1500. I'd like to recommend there might need to be a liaison statement to IEEE to fix this. Ron: I agree Jared Mauch: Ron: Jared: Jen: We need to say it's not just filtering but also other things like not generating. DNSSEC is a big problem and we need recommendations to network operators to make that work. Ron: OK Juliusz: I think this is a good document, but do have some comments. It's not clear who you are talking to -- it may be good to target application developers. Ron: That might be good. Christian Huitema: You don't mention that fragmented packets don't necessarily arrive in the same order. Because they don't have the same port numbers, they don't go in same buckets and can arrive in any order. Geoff Huston: In theory one should never see fragmented TCP packets. DNS is a problem but will fix after a while. Why not fall back to TCP? In DNSOPS WG, they should be looking at draft that proposes solution -- how to make flip to fragmentation recovery quicker. If you must fragment, recover more quickly. Assume breakage and recover. Jared Mauch: The problem I see is router vendors tend to do this poorly. We have rate limiters that cause these packets not to make it back. Our end customers have high performance expectations. Regardless of app. Performance is gating thing. Ron: My take-away is we need to have recovery mechanism. Lorenzo: We need to look at other ways to solve this. Ron: We need to work this problem. Jinmei: some quick comments regarding DNS: the DNS community is now more interested in using TCP (although DNS over UDP won't disappear). There's a proposal of application layer fragmentation of DNS several years ago (and it's now dead). Will send a link. Ron: Thank you ----------------------------- Jordi Palet Martinez presented NAT64 Deployment Guidelines in Operator and Enterprise Networks https://datatracker.ietf.org/meeting/101/materials/slides-101-v6ops-nat64-deployment-guidelines-in-operator-and-enterprise-networks-00 Jordi Palet Martinez presented IPv6 Point to Point Links https://datatracker.ietf.org/meeting/101/materials/slides-101-v6ops-ipv6-point-to-point-links-00 Jordi Palet Martinez presented Transition Requirements for IPv6 CE Routers to support IPv4 as a Service https://datatracker.ietf.org/meeting/101/materials/slides-101-v6ops-transition-requirements-for-ipv6-ce-routers-to-support-ipv4-as-a-service-00 Jen: I looked at NAT64 doc and got confused. It scares people away. Regarding DNSSEC problem, maybe tell people if you care about security then you must make it work. Barbara: Need to narrow list of transition technologies more, if possible. Lee: I have a list of who is running what and will post. Jordi: In old survey, MAPT, MAPE, Lightweight were little used. But use is increasing. Fred: We are at top of the hour. Take further discussion to the list.