IETF IPv6 Maintenance Working Group Meeting Wednesday, 5 December, 2007, 0900-1130 The meeting was called to order by co-chairs Brian Haberman and Bob Hinden. Karen O'Donoghue (with input from Ed Jankiewicz) took minutes and Suresh Krishnan acted as Jabber scribe. The blue sheets were distributed, and the agenda was presented with no additions. During the agenda discussion, Rosalea Roberts reminded the working group of the draft-narten-ipv6-3177bis-48boundary-03 discussion currently ongoing in v6ops and encouraged this working group to follow and participate in that discussion. Node Requirements Update (Brian Haberman for John Loughney) ================================================= Document: RFC 4294 IPv6 Node Requirements Slides: IETF70-6MAN-node-reqs-update.pdf Brian Haberman, speaking for the unavoidably absent John Loughney, discussed the question of updating RFC 4294. There are basically two questions. Should we do the work and if yes, how should we proceed? Several individuals indicated that the update should be done and there was no opposition. Based on the consensus that the work should be done, Brian further asked how to proceed. Three options were identified: 1. Simple update to cover base RFCs 2. #1 AND adding additional RFCs 3. #2 AND restructuring The ensuing discussion primarily focused on options 2 and 3 with most speakers supporting option 2. A recurring theme was the need for a definition of the minimal requirements for interoperability with organizations responsible for deployment (e.g. DoD and NIST) specifying additional deployment specific requirements. Others indicated that the current document currently contains very few MUSTs as it is. The use of conditional language to specify requirements for subsets of functionality was discussed. Brian Haberman requested that participants continue the discussion on the mailing list. IKEv2 issues (Pasi Eronen) ===================== Draft: draft-eronen-ipsec-ikev2-ipv6-config-01.txt Slides: IETF70-6MAN-ipv6-ikev2.pdf This draft addresses an issue with IPv6 configuration in IKEv2 for remote access VPNs. There was general agreement that IKEv2 is broken and this works needs to be done. Discussion focused on whether the link model utilized was representative enough and potential scalability issues. Yari Arkko indicated that he liked the draft and supported the effort. A change is needed and there is a small window of opportunity to make this change. Bob Hinden asked if it was appropriate for the 6MAN WG to do the work. Yari prefers for the security ADs to sponsor the work with strong participation from this WG and will take the action to initiate a conversation with them. Reserved IIDs (Suresh Krishnan) ======================== Document: draft-krishnan-ipv6-reserved-iids-02.txt Slides: IETF70-6MAN-reserved-iids.pdf Suresh presented this proposal to create a centralized repository for reserved interface identifiers. Brian Haberman pointed out that we already do this for the multicast address space (RFC 3307). No one opposed the adoption of this as a working group document. DHCP options in RAs (Suresh Krishnan) ================================ Document: draft-krishnan-intarea-ra-dhcp-00.txt Slides: IETF70-6MAN-dhc-options-in-ra.pdf Suresh presented a draft that defines a generic neighbor discovery option for carry DHCP options over IPv6 Router Advertisements. The rationale included a desire to avoid parallel definition and duplicate standardization as well as the potential to share code. During the discussion phase, several individuals addressed concerns with the proposal. Alain Durand stated that this draft had been discussed yesterday in the DHCP WG and the sense of the DHC meeting was that it was a bad idea. Bob Hinden pointed out that IF this goes forward the WG would need to classify current options. There is probably a very small set of options that this would be appropriate for. Erik Nordmark stated that this creates more problems than it solves. The more places we put this kind of information, the more work we have to do to resolve what happens when you get contradictory information from another method. We also need to be very specific about the lifetime of this information. Dave Thaler pointed out that without this proposal, anyone who wants to put the same information in RAs and DHCP must come to the IETF for review and approval. Ralph Droms agreed that by doing this we eliminate possible oversight. One individual did indicate that this was a good idea, but we need to be very careful about what information is allowed to be carried. Yari Arkko concluded the discussion by stating that he did not support this draft. Address Selection (2 presentations) ========================== Document: draft-arifumi-6man-addr-select-sol-00.txt Slides: IETF70-6MAN-arifumi-addr-select.pdf Arifumi Matsumoto presented the above draft. Erik Nordmark stated that the document talks about a node with multiple interfaces. How would you deal with the situation where you got multiple responses? The chairs asked the room who had read the document and who thought it was ready for to be adopted as a working group item. There was a small set of people answering yes to both. Further discussion was directed to the list. Document: draft-fujikawa-ipv6-src-addr-selection-01.txt Slides: IETF70-6MAN-fujikawa-src-addr-selection.pdf Fujikawa Kenji presented the above draft. Time ran out and the chairs requested that the WG read the document and take the discussion to the mailing list. On-link issues & 2461bis (Hemant Singh) ============================== Document: draft-wbeebee-on-link-and-off-link-determination-00.txt Document: draft-wbeebee-nd-implementation-problems-00.txt Slides: IETF70-6MAN-offlink.pdf Hemant Singh presented a discussion on the on- and off-link issues with RFC 4861. Hosts in aggregated routed networks are offlink. There is data forwarding confusion on IPv6 hosts behind modems in aggregated routed networks. RFC 4861 is ambiguous on how to configure RA on a router to signal off-link. The author wishes to clear up ambiguity in the specification. Erik Nordmark said this sounds like an implemention bug. Bob Hinden indicated that more discussion and analysis are needed before we can adopt as a working group item. He directed discussion to the mailing list. Dickson draft (Brian Dickson) ====================== Document: draft-dickson-v6man-new-autoconf-00.txt Slides: IETF70-6MAN-New-Autoconf.pdf Brian Dickson presented a proposal that is motivated by looking at wide and timely IPv6 deployment and issues of scaling in a mixed IPv4 and IPv6 network. The draft proposes the use of prefix lengths other than 64 bits in certain circumstances. The purpose of this draft is to foster discussion and collect some analysis for consideration. Brian wants to know if we can make a small change to improve scalability? An animated discussion followed Brian's presentation. Tony Hain stated that he thought this was a really bad idea. It breaks current operations, the fundamental premise that it helps scaling is flawed, and the proposal is ten years too late. Marla Azinger stated that the document was too long and she was disappointed that the author didn't get to the point thus obscuring the message. She felt we couldn't decide today because more work is required to clarify the arguments and proposal. Alain Durand voiced some sympathy with the idea that maybe /64 is not the right thing; however, he felt that we need to separate the policy and technical discussions. Margaret Wasserman pointed that while we could decide whether or not this is good technically we are not in a position to force a policy change. Another speaker felt the analysis was good, but the idea was bad. Erik Nordmark seconded the concern that the timing is wrong (it is either 10 years too late or 20 years too early). He also is not concerned about current IPv6 address consumption rate. Another individual pointed out that we don't fully appreciate the impacts of this on other organizations. Brian Dickson concluded with the comment that making these changes later would require re-prefixing and re-numbering. A parallel move towards incremental changes would not impede deployment. The chairs asked the room whether the work should continue. Many hands said no and a few indicated yes. ULA-C Analysis (Margaret Wasserman) ============================= Document: draft-mrw-6man-ulac-analysis-01.txt Slides: IETF70-6MAN-ULAC-Analysis.pdf Margaret Wasserman presented her draft on an analysis of Centrally Assigned Unique Local Addresses (ULA-Cs). The draft discusses the motivation, costs, and some of the arguments for and against ULA-Cs in an attempt to help the IETF IPv6 community reach consensus on this issue. The questions for the WG are: 1) Should some kind of centrally-assigned ULA be available; and if so 2) should they be defined in the IETF? Bob Hinden's draft was produced with the motivation that he wanted a good example of what it could look like if the IETF wanted to do this. He is not a strong proponent. The world has gotten used to local addresses, and he doesn't want the lack of net 10 type addresses to slow down v6 deployment. Kurt Lindquist felt that ULA-Cs create a sixth RIR and is a bad idea. Alain Durand indicated that this is a technology versus policy issue, and he's not sure where it should be discussed. Tony Hain indicated that we are not creating a registry. Instead, we are creating an instruction to the registry community. Someone will create this and do it anyway if the IETF doesn't. Dave Thaler indicated that the benefits don't outweigh the costs. The chairs asked that folks read the draft and comment on mailing list. The meeting adjourned.