Mobile Ad hoc Networks (MANET) Minutes of meeting Meeting : IETF 90 Tuesday July 22, 2014 Time : 1420-1620 Afternoon Session II Location : Manitoba Chairs : Joseph Macker Stan Ratliff Secretary : Ulrich Herberg Jabber : xmpp:manet@jabber.ietf.org Scribe : Rick Taylor Minutes : Ronald in 't Velt Audio archive : http://www.ietf.org/audio/ietf90/ietf90-manitoba-20140722-1420-pm2.mp3 URLs : http://www.ietf.org/html.charters/manet-charter.html http://tools.ietf.org/wg/manet/ ========================================================= o Administrivia - Mailing list: manet@ietf.org - Scribe(s) - Blue Sheets - State your name at the microphone - IPR Ulrich: be aware of the "Note well". Proxy chairs: Ulrich Herberg, Justin Dean o Bash the Agenda [5] No changes. One minute time slot for Interop announcement to be made at the end of the meeting. WG Status/Overview - Recap of Documents/Status * OLSRv2 "cluster" of RFCs published * Active WG drafts: - AODVv2 new revision - DLEP no new revision Thomas Clausen: worried about this, no new version since two meetings. Should WG be worried, take corrective action? Justin Dean: There was a Design Team meeting at IETF-89, should have resulted in a new version, has not happened. Rick Taylor: Outcome of that meeting was summarized on the MANET ML. Thomas Clausen: Design Team is larger than the author set. Add authors? Current authorship is perhaps running out of steam. Rick Taylor: have had correspondence with Stan Ratliff. A new draft version is being worked on. Progress is slow, however, due to lack of time. Thomas suggests again that the WG should decide to add authors. - New I-D by Chris Dearlove: Identity-Based Signatures (IBS) for MANET Routing Protocols o Ulrich will present it on behalf of Chris at this meeting - New WG draft NHDP Optimization - New version of DAT metric I-D, by Henning Rogge. Thomas Clausen: Go for WGLC soon. Found only a few nits. This draft should at least be in the hands of the AD before IETF-91. Justin Dean: I have implemented the previous version of this draft, found it straightforward. Agree with Thomas. Remaining nits can be addressed during WGLC. - OLSRv2 Management Snapshot Draft o Will be presented by Ulrich o Want to ask for WGLC - OLSRv2 Multi-topology I-D o Has passed WGLC o Is waiting for Chairs' write-up - SMF MIB I-D o Waiting for revised I-D o Question about status on ML. Bob Cole responded that the appropriate changes just need to be made. There are no major issues. - Report MIB I-D. o Draft expired for a year and a half. Appears stalled. o Offer from WG participants to help Ulrich Herberg: Bob Cole indicated that he considered changing it to 'Experimental' as well as changing the structure. Thomas Clausen: We need a resolution on this. Want to reference this from one of my own I-Ds, but only if standards track. (Otherwise it would be a downref). There has not been a new version for two years. Maybe we should declare this document dead. Might as well make that official. Comment on Jabber from Abdussalam Baryun: Agenda: present WG drafts before individual I-Ds. Ulrich Herberg: Too late, agenda has been set. All drafts can be presented if presenters stick to their allocated time slots. Rick Taylor: Is working on the MIBs still relevant in the face of the apparent transition within the IETF to Yang and Netconf? Thomas Clausen: recently asked the same question to an AD. Reply: if you have started with MIBs, continue. For new work, don't do MIBs for writable date structures. Adrian Farrel (AD): Yes, it is a transition. It would be painful to have to support both. WG might like to look at transitioning MIBs to Yang, for the benefit of new implementations. It is a long process, however. Justin Dean: I know that Bob Cole and his team are aware of Yang, have some expertise there. Adrian Farrel: WG may want to appoint someone to help Bob Cole with the Report MIB. Or decide to abandon it. In any case, make a decision, don't let it fester. Ulrich Herberg: W.r.t. IESG statement (https://www.ietf.org/iesg/statement/writable-mib-module.html): The Report MIB is not a writable MIB module. * Individual I-Ds: Thomas Clausen: On this list: draft-baccelli-manet-multihop-communication- 03. This document has a bit of a turbulent history. Last presented in IntArea at IETF-89. Suggest that this WG takes a look at it. Ulrich Herberg: Concluding WG status. With the publication of the OLSRv2 cluster of RFCs, the WG has completed one of the two major Work Items on its Charter, i.e., the Proactive MANET Protocol. * I-D Presentations: * draft-ietf-manet-olsrv2-management-snapshot: Herberg - (see slides) - How, what, why OLSRv2 MANETs are to be managed - Status update since WG adoption. Feedback from OPSAWG resulted in new section on typical (management) communications patterns. - Feedback from WG needed. Go for WGLC 1 week after IETF-90 * draft-dearlove-manet-ibs: Herberg - (see slides) - Ulrich presenting on behalf of Christopher Dearlove. - Now has been WG adopted. - Justin Dean: Christopher asked for others to use his appendix to generate key as a way to verify the algorithm. * draft-yi-manet-smf-sec-threats (and other security work): Clausen - (see slides) - Thomas Clausen on behalf of Jiazi Yi. - What's new relative to threats to NHDP already identified. - Related: draft-clausen-manet-olsrv2-sec-threats-00, too recent to be discussed today. Thomas Clausen asks the WG to read the I-D and comment. - Request WG adoption - Rick Taylor: What is the status of SMF? - Thomas Clausen: Experimental RFC. - Not many people have read the draft. Hum on smf-sec-threats being a good idea: hum in favor. Same for olsrv2-sec-threats. - Adoption request to be taken to the ML. * draft-ietf-manet-olsrv2-multitopology: Clausen - (see slides) - Aimed at Experimental RFC - WGLC completed. WG chairs requested to assign doc shepherd. - Ulrich Herberg: I volunteer to be the shepherd. * draft-dearlove-manet-nhdp-optimization: Clausen - (see slides) - Now WG adopted: draft-ietf-manet-nhdp-optimization-00 - Next steps: Find out how to update NHDP MIB module. Need more WG reviews. Would like to ask for WGLC soon (but not now) and have this with RFC Editor before IETF-91 * draft-yi-manet-olsrv2-multipath: Clausen (proxying Yi) - (see slides) - Uses source routing for packet forwarding - Aiming for Experimental - Asking for WG adoption. - Adrian Farrel: why is this specifically MANET? - Thomas Clausen: Because information is being added to TC messages, therefore OLSRv2. Could be done with another IGP as well. Would however like to deflect to the authors on this. Probably also to exploit multiple wireless interfaces on boxes that have them. - Jiazi Yi on Jabber: more interesting in MANET, because more unreliable. We found that multipath has more performance gain in dynamic and lossy scenarios. - Adrian Farrel: Is loose source routing done with standard IP source routing headers? - Thomas Clausen: Yes, in my code it is. - Adrian Farrel: we may want to have that debate with IntArea. Also look at what the SPRING WG is doing. - Adrian Farrel: Does changing the metrics mean changing those in the local representation of the graph? - Thomas: Yes. - Adrian: do all nodes have to do this in the same way? - Thomas: No, this is a completely local decision. - Justin Dean: who read the draft? (2 people). Taking a hum on the idea of adopting this I-D as a WG Draft: several in favor, one against. Justin Dean: needs more discussion on the ML. - Charlie Perkins: agree, more discussion needed. Is this for OLSRv2 only? What about other MANET protocols. - Thomas Clausen: I have no answer to that. - Henning Rogge (Jabber): OLSRv2 is the only MANET protocol using Dijkstra. - Abdussalam Baryun (Jabber): It is not right to ask the WG to review many documents in a short time. - Jiazi Yi: some multipath extensions for AODV exist, not necessarily using Dijkstra. - Ulrich Herberg: Humming is not voting. Need to hear arguments pro and con. * draft-gerla-manet-odmrp: Clausen (proxying Colin de Verdiere) - (see slides) - Updated versions (-03 and -04), incorporating feedback from WG - Aiming for Experimental RFC - Next steps: ask for WG adoption. - Justin Dean: Gave feedback on previous version. As for WG adoption, not sure whether charter covers this. Together with all the current OLSRv2 extensions, doing multiple multicast protocols could be too much for the WG. - Thomas Clausen: multicast is within Charter. - Thomas Clausen: as a personal view, I think we need group oriented multicast. - Vince Park: lots of experimental stuff. Great, but not reflected in carter. Should give more thought to next steps. - Charlie Perkins: M-AODV, presented many years ago. Thought multicast was out of scope. Other proposals for multicast may not have surfaced because of that. Even if ODMRP is published as experimental, it needs an applicability statement. - Ulrich Herberg: (reads relevant charter text: "to all participating MANET nodes"). - Toerless Eckert: It would be good to present these multicast protocols also in PIM WG. It is called 'PIM' for historical reasons, but now encompasses other multicast routing work as well. Make use of expertise there. - Abdussalam Baryun (Jabber): Agree with Justin on need to limit our work, agree with Charlie: standards first. - Thomas Clausen: agree on making use of outside expertise, do not see the charter as excluding this, I have a need for multicast. Experimental status because we do not have the operational experience yet. Need to experiment. Would like to see work from NRL brought to the WG as well. Having some competition is good. - Rick Taylor: agrees there is a need for multicast from the users of MANET protocols. - Ulrich Herberg: Let's continue this discussion on the ML. TC wraps up his presentations with a "Collective Next Steps" slide. * draft-taylor-manet-l3-dlep: Taylor - (see slides) - Individual draft - Layer 3 extensions to DLEP. Two minimal optional mechanisms. - Secondary goal: test case for extension mechanism. - DLEP assumes bridging radio modems. Vendors of L3 radios want DLEP too. - Henning Rogge (Jabber): How do L3 radios do multicast? - Rick Taylor: good question! By setting up groups, etc. This needs more work. - To Do: Not yet ready for WGLC. (Rick Taylor: "Miles off"). - Rick Taylor asks WG participants to read the draft and comment. - Toerless Eckert: More analysis is needed. What would routers on 'router' side of DLEP (as opposed to the L3 modems) be willing to do. Tunneling from the DLEP routers is no good, been there years ago with satellite modems, seems you do need something BGP-like. You need to talk to people that want to deploy this stuff. - Rick Taylor: We are. Small networks, BGP seems to heavy. - John Dowdell: too many AS numbers would be needed. - Thomas Clausen: what is the dependence on DLEP Core? Rick Taylor: This is an extension. Don't believe we are dragging a lot of DLEP functionality along that is not needed for L3 modems. * draft-snyder-bfd-proxy-connections-monitored-links: Snyder - (see slides) - Satellite modem vendor. Looked at DLEP. Does not solve my problem. Then looked at BFD. - Toerless Eckert: trying to understand, 1000's of modems, BFD between all of them. - Brian Snyder: not a mesh, star topology. - Rick Taylor: DLEP Rationale doc needed, because this is based on assumptions on DLEP that have changed in the meantime. * draft-ietf-manet-aodvv2: Perkins - (see slides) - Added more algorithmic description of the protocol as Appendix A, complementing the text oriented description in the main body of the document. - Thomas Clausen: what's the goal of the appendix? - Charlie Perkins: the goal is to have a representation in this form for people who want it. Main body is more explanatory. - Thomas Clausen: There is a risk of ambiguities. Which one is authoritative: main body or appendix? - Charlie Perkins: the main body is normative. - Thomas Clausen: If Appendix A is not normative, then I suggest to remove it. - Terminology changes - Made clear that multiple addresses per interface are allowed. - Thomas Clausen: Multiple interfaces, multiple addresses. Should be very careful here. Learned from RFC 6130, that this is hard to get right. I think what is in the draft now is not correct. E.g. blacklisted router, versus blacklisted address. - Rick Taylor: Been there with a different protocol. You should blacklist the link, not the router. - Charlie Perkins: I think we can solve this. - Issue tracker. Charlie Perkins: Please use it more. - Thomas Clausen: concerned about relation to RFC5444, RFC5498. The architecture as laid down in these documents should be respected. A second concern is the number of places in the document where the choice between several approaches is left open. This is not helping interoperability. - Rick Taylor: There are RFC5444 experts/authors in the WG. They could try to help Charlie Perkins. I find some of the criticism a bit harsh. - Thomas Clausen: Disagree. Meeting ends, time has run out.