Meeting minutes Softwire WG 16:17:00 2013/11/5, Georgia B, Park Hyatt, Vancouver, BC, Canada Softwire WG IETF 88 Tuesday, Noverber 5, 2013 1610-1840 Afternoon Session Minute takers: Xiaohong Deng, ISOC fellow to IETF --- Administrative and WG document status Chairs Suresh, Chair: no new documents. Keep the meeting short and focus on making progress of the existing works. Cui, Chair: Agenda Bashing and Agenda approved. --- 01. MAP-E update Satoru Matsushima https://datatracker.ietf.org/doc/draft-ietf-softwire-map/ Suresh: will do a short WG last call again before sending it off --- 02. A Proposal for the Softwire WG Ian Farrer Ian, Deutsche Telekom, presented DHCP provisioning method for MAP-E,MAP-T and lw406 and MAP three solutions and explained what's going on off-line (actually via on line :)) since IETF87. Three containers built out of common sub-options for MAP-E, MAP-T and lw4o6 respectively. Explained Relations to other provisioning: PCP, DHCPv4: doesn't preclude or prevent them, but provides a foundation. Ask people if agree to use OPTION_S46 for both provisioning MAP-E and lw4o6? Alain, Juniper: we've been here for long time, so move on with something work-able. There are things that I like and that I don't, but the questions is if I can live with it Ida, Rogers Com: I'd like it to go fast so I can implement it. Suresh: What solution you want implement? Ida: MAP-T fits my implementation plans. Suresh: WG does not prevent any of them including MAP-T from going forward; Progress depends on members' effort, so if you concern it, put effort on it, for example, implement it. It's everyone's participation that makes it goes faster Andrew, Cisco: only two containers appear there, I want to have all three. Roberto, Cisco: we have SPs who would like all solutions available. Alain: trying to boil the ocean is why we were not making progress fast. Suggest to stick to what the WG consensus and deliver the whole the package we agreed in WG. Suresh: Yes. MAP-T is not in the package. Do you agree the provisioning proposal here? Alain: If it is explicitly specified in the document that MAP-T is implementable then I don't object. Suresh: Yes, it will be specified clearly in the document. CongXiao: MAP-T has the same level of maturity as MAP-E. Decision based on tossing coins doesn't look fair. Suresh: Decision based on tossing coin was if the WG treat MAP-T and MAP-E together, not the choose between the two. No one is making claiming the differences of the maturity. WG's consensus was that MAP-E should go first. WG doesn't hold on the progress of MAP-T or others either, it just doesn't send them at the same time with MAP-E. Sheng: what's the plan for 4rd. Does it require 4rd using map provisioning? Suresh: No. Sheng: I'm fine by that. Alain: when will the documents be sent to IESG? Suresh: after MAP-E, all three documents will be sent together to IESG. --- 03. Unified CPE provisioning Simon Perreault https://datatracker.ietf.org/doc/draft-ietf-softwire-map-dhcp/ https://datatracker.ietf.org/doc/draft-ietf-softwire-unified-cpe/ Simon, Viagenie presented Unified CPE: Explained what happened off line among authors and those who showed concerns before. The principle is: Presence of options depending on request, content of options doesn't depend on the request. Sub-options are open to discuss. The principle is based on the philosophy of making CPE as most unified as possible. No Negotiation between DHCP servers and CPE. It's up to CPE when multiple options provisioned for backward compatibility. Unified CPE tries to unify MAP(both E and T), LW4o6 and DS-Lite all protocols together, while DHCP provisioning draft deals with MAP and LW4o6 only. Roboto, Cisco: when you say MAP you mean both E and T? Simon:Yes. Cong: if people need other options for other solutions, how DHCP provisioning behave? Ian: Container option is the way to link things all sub-options together. Suresh: we'll try to understand and answer your question here: what else DHCP options you want here? Cong: for example, Ds-Lite. Simon: no need to provisioning DS-Lite. DS-Lite is provisioned with other existing means. Suresh: Unified CPE do more than DHCP provisioning. DHCP provisioning draft doesn't concern DS-Lite . Bernie: If you want v4 stuff, do DHCPv4 Suresh: data path for Unified CPE has more than what containers do Qi: statement in map-dhcp-05 about lw4o6 is subset of MAP-E Suresh: It will be gone. Qi: next version remove it? Suresh: Yes Qi: As softwire-dhcp draft, it should use neutral word for 3 mechanisms, may need rewording Suresh: You can open an issue. Title already changed. Do you agree with the proposal? Qi: OK