Mobile Ad hoc Networks (MANET) Minutes Meeting : IETF 85 Wednesday November 7, 2012 Time : 1300-1430 Afternoon Session I Location : Salon A Chairs : Joseph Macker Stan Ratliff Secretary : Ulrich Herberg Jabber : xmpp:manet@jabber.ietf.org Audiocast : https://www.ietf.org/audio/ietf85/ietf85-salona-20121107-1300-pm1.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 o Bash the Agenda Agenda was accepted. o WG Status/Overview: Macker/Ratliff - Recap of Documents/Status - Announcements Working Group chairs presented document status for working group documents. Some discussion regarding comments on olsrv2-metrics document, late arriving comments. Chris Dearlove: Would be nice if comments were provided for end of WGLC. Thomas Clausen: Eager to see progress on this document. Charlie Perkins: How to reference reduced relay sets as normative reference was raised on the list. Joe Macker: I provided answer on list (publications with reduced relay set algorithm psuedo-code), was that acceptable? Charlie Perkins: Can use the citation provided there for that, but also the same question comes up for metrics, so how would OLSRv2 metric design impact other protocols? Thomas Clausen: This document is specific to OLSRv2 and not something to cite for other (e.g. reactive protocols) purposes. Joe Macker: need to move on. o Reactive MANET protocol: Perkins/Herberg - draft-clausen-lln-loadng - Ulrich Herberg, presenter Contents of presentation: 10 authors with diverse background and companies, LOADng is derived from AODV, well-known and studied algorithm, slim base feature set, reduced complexity, modular (similar to OLSRv2), options avoided unless proved to be useful in a general MANET case, to avoid interop problems, embedded in MANET architecture of RFC5444,5498,5148,6621,6622, support of optimized flooding, discovery of bi-directional neighbors, messages constructed of mostly immutable fields (end-to-end security), multiple interfaces per router, extensible Several deployments (e.g. one with 2000 nodes from ERDF), three interop events with four interoperable implementations, MIB document available Ulrich Herberg: review of this and AODV2 is encouraged. - draft-ietf-manet-dymo - Charlie Perkins, presenter Charlie Perkins note: Slides may not be complete. Contents: Plan for LOADng compatibility: AODVv2 base protocol is already close, few minor details, I have a pretty clear understanding of goals, input from LOADng team welcome, as of now I am prohibited from using LOADng text. Specification revisions in 2012: March revision was done in extraordinary hurry to satisfy WG editorship expectation, plan for a way forward: none was found, October: document merge was not happening, soon: a "non-hasty", quality product Editorial responsibility: responsive to WG input, manage document text revision, manage document text revision, manage ID submission, prodding for mailing list input, resolution via IETF tools issue tracker, fair to all parties, expedited completion of charter goals Chris Dearlove: Not a DYMO or LOADng author. Made a proposal on the list to differentiate between documents and protocols. Opinion is LOADng document is a better starting point. Charlie Perkins: Are you willing to revisit opinion upon dymo-24 release? Dearlove: Yes, I will review new documents when they come out. Jiazi Yi: LOADng is a better starting point. Don't see compatibility between DYMO and LOADng. Charlie Perkins: A lot of instability with document is because of changes needed for RFC5444 compliance. There is a comment in the LOADng spec that it is not yet quite done yet, so how can you claim interoperability maturity? Stan Ratliff: You stated that DYMO and LOADng are not compatible. That is a fundamental problem. I have understood that they were fundamentally similar. Justin Dean: Jiazi is mostly talking about packet format. Joe Macker: You are stating they are not _currently_ compatible, but could be with a little work? Jiazi: not sure. Henning: (to AODV2) "Does LOADng compatibility mean you want to participate in interop tests with the existing implementations?" Perkins: In a converged world where people are acting to achieve the same goals, and if LOAdng is a matter of adherence to certain features that are specified in the AODVv2 document, then any interop tests also suffices for interop tests for AODVv2. Abdussalam: Are there results of performance evaluation or references I can read for LOADng? Justin: Anything relating to AODV is applicable for LOADng. Joe Macker: Lots of papers available Carsten Bormann: We have a working group document and a document that is submitted that no modification allowed? How did we get here? Stan Ratliff: There is not a short answer but will talk about it off-line. As working group participant, text in LOADng banning derivative works should be removed. Thomas Clausen: Don't know what it takes for one protocol to be "compatible" with another as opposed to two implementation interoperable. Answering Carsten, the DYMO document had stalled. With respect to the copyright notice, it can be changed. Carsten Bormann: When you say the copyright statement _may_ be removed... Thomas Clausen: As a LOADng author, would push for it to be removed. Joe: I also support removing the copyright suggestion. JP Vasseur: To Joe Macker, do you believe that LOADng is not LLN applicable? Joe Macker: The IETF taxonomy is not perfectly divisible ... these are AODV protocols and are OK for MANET. Want to avoid LLN performance comparisons besides MANET. But we need to be aware, of issues across both ROLL / MANET. JP Vasseur: Applicability statement needs to be clear on this. Joe Macker: Working group chair direction is to not state LLN JP Vasseur: The deployment mentioned is LLN and 2000 node deployment. Joe Macker: The decision to proceed is based on AODV design JP Vasseur: I support AODVv2 and that the document is maturing rapidly as a superset with options. Charlie Perkins: Answering question regarding protocol compatibility: good engineering sense that the protocol specifications work the same ... Joydeep Tripathi: LOADng is a subset of DYMO ... there are DYMO features that would be useful to more general MANET deployment. For example, the precursor list option of DYMO. Joe Macker: To clarify, you like some of the DYMO options like precursor lists, etc. Joydeep Tripathi: Can restrict DYMO to be more like LOADng. Thomas Clausen: Precursor list can be beneficial in some cases and bad in others. It is an option that can be used in some deployments. Decision to use precursor list in implementation doesn't affect signaling and thus interoperability. So, vendors can choose to implement it or not ... WG Chairs: These technical points should be discussed on the mailing list (It's a long, serious technical discussion). There are multiple options like these that should be listed and discussed on the mailing list. Thomas Clausen: Charlie wants to DYMO to look, taste, feel smell like LOADng ... if you want to move in this direction, let's start with the LOADng draft. Stan Ratliff: It is reasonable to read and see that one is a superset of the other. They are both AODV based ... is it making AODV2 look like DYMO or Abdussalam: Is there a possibility of merging? Charlie Perkins: DSR and AODV were merged to make DYMO as an example from the past. Joe Macker: We have to look at _all_ the issues that a standards track document needs to address, including SECURITY, etc. Be careful about throwing in too many options that possibly create other problems. JP Vasseur: LOADng makes some assumptions on the traffic profile ... Thomas Clausen: Not sure I understood the question, but didn't have assumptions. Justin Dean: Encourage everyone to read both of the documents. I think the 3 options presented aren't the only three, but think the working group can make this into one document. Stan Ratliff: Would like to see consensus on way forward. My preference is to remove the work item from the MANET charter. Carsten Bormann: We have two documents, and we would like to merge them ... so we are discussing which document to start from. Documents are malleable, so what you start from is second order. Having the right editor is key and working group chairs can decide this. Joe Macker: Need author energy, etc to be able to maintain consensus Carsten Bormann: Don't need consensus to choose an editorial. Do we have an editor can move this forward. Chris Dearlove: the documents are different quality, etc with respect to presentation styles Ulrich Herberg: Do we have a single editor, there were/are personal issues, history involved here. A possible way forward would be to select two editors from the industrially deployed authors. JP Vasseur: Given we have a working group document, why can't we go ahead with it [AODV2] since it is a superset of the other. Charlie Perkins: For a document comparison, wait for dymo-24 release ... Stan Ratliff: I see JP's point and this is what we are trying to get consensus. Haven't heard consensus yet? Stan Ratliff: There has been quite a bit of work off the list. Alex Petrascu: This is a hot issue and many are watching. Should discuss what sort of tools the authors are using and be considered, such as the IETF issues tracker, etc Thomas Clausen: Can provide accounts on LOADng issue tracker. Why did we go with LOADng instead of fixing DYMO ...? As an individual, I have embraced inviting all and any to participate, including Charlie Perkins. Let the people that are deploying this and have an interest in the technology to be editors (perhaps a pair chosen from competing parties) Carsten Bormann: Recommend that you give those suggested pairs to chair, open the document and let them decide. Thomas Clausen: OK - authors present at meeting OK with lifting IPR restriction (2 more authors need to be asked) Joe Macker: To Thomas: need to have decision on freeing the document, suggested editors, etc soon. Stan Ratliff: Consensus evolve around approach to merge documents, and chairs need to balance that with "personal, point of view" issues that have popped to life. Adrian: Not just get a verbal, but a get a new draft with IPR removed. Charlie: If next document doesn't have IPR restriction, can I use the text in DYMO (e.g. dymo-24) o OLSRv2-MIB Discussion: Herberg - draft-ietf-manet-olsrv2-mib - draft-ietf-manet-smf-mib - draft-ietf-manet-report-mib Ulrich Herberg: I believe OLSRv2-MIB is ready for WG LC. o MANET management use case draft: Herberg Ulrich provides brief status on MIB and Management status. "Use Case" draft was required for MANET MIBs to proceed. Ulrich has a plan to go forward with that, forming a design team to produce a draft. o DLEP: Ratliff - draft-ietf-manet-dlep David Ward: He and colleagues had submitted a draft describing the set of metrics that you might want to get from a radio. Is the group interested in this sort of draft? Stan Ratliff: Some experimental TLV space is allocated for better metric set. I don't want to "spin" for years on specific metrics and delay DLEP draft. I had some discussions with radio vendors saying they will implement it if it gets past working group last call. Thomas Clausen: 03 was better than 02 ... RFC5444 format not good/needed for DLEP (after having implemented DLEP). Stateful session-based vs. session-less ... confirms the design decisions made there (stateful). Can the two documents (metrics for DLEP) go forward in parallel (R2RI document) with appropriate IANA registries for the types? Rick Taylor: R2RI is a good and useful document. Henning: When will dlep-04 be published? Stan: In 3-4 weeks. Rick Taylor: Suggested metrics ... agree RFC5444 is unnecessary. William Ivancic: Are you considering a DLEP-Lite such as modemLPA? Stan Ratliff: Yes - we're trying to improve the spec any way we can. Chris Dearlove: on OLSRv2: Impact of Security AD discuss on philosophy of how we should be doing security ... should be discussed on the list. Teco Boot: On DLEP-Lite, there may be crypto boundary issues that require it, i.e. to be stateless. Chris Dearlove: Excellent point from Teco Boot. no time left for: o OLSRv2 Metrics Rationale: Clausen - draft-ietf-manet-olsrv2-metrics-rationale o Security Documents: Yi - draft-ietf-manet-nhdp-sec-threats o Open Microphone - Discussion, Related Work & Announcements - Interops, Implementations, Other Items