Clone wiki

meetings / 140328_webex

Minutes Webex 28 March 2014, 6TiSCH WG

Note: timestamps in PDT.

Taking notes (using Etherpad)

  1. Xavi Vilajosana
  2. Pascal Thubert
  3. Thomas Watteyne

Present (alphabetically)

  1. Diego Dujovne
  2. Dominique Barthel
  3. Giuseppe Piro
  4. Kazushi Muraoka
  5. Maria Rita Palattella
  6. Michael Behringer
  7. Nicola Accettura
  8. Pascal Thubert
  9. Pat Kinney
  10. Patrick Wetterwald
  11. Pouria Zand
  12. Qin Wang
  13. Raghuram Sudhaakar
  14. Thomas Watteyne
  15. Xavi Vilajosana

Recording

Slides

Action Items

  • Raghuram and Pouria to publish new version of CoAP draft.
  • Qin to announce 6top-to-6top call on ML, possibly creating Doodle to agree on date, and as Pascal to set up webex.
  • Patrick to send pointers to the ML where the terminology is not clear.

Agenda

  • Administrivia [7min]
    • Approval agenda
    • Approval minutes last call
    • publication IETF89 minutes
    • publication draft-ietf-6tisch-6top-interface-00
  • Webex reorganization [10min]
    • proposal
    • discussion
  • 6top-2-6top protocol [30min]
    • overview discussions
    • proposal
  • Lessons learnt minimal draft [time permitting]
  • reminder next calls [2min]
    • 6TiSCH webex
    • OTF call
    • security call
  • AOB [1min]

Minutes

  • [08.05] Meeting starts
    • recording starts
  • Administrivia
  • [08.11] Webex reorganization
    • Change webex periodicity to be able to support parallel calls.
    • Turn weekly call into a bi-weekly call.
    • Have (smaller) calls for specific topics during the week.
    • Goal is to encourage more discussion.
    • Ask for more usage of the ML.
    • Move individual emails to the ML so this is a general forum.
    • Webex can be setup on demand to support group discussions
    • Requirements: note well, minutes, announce the calls, publish minutes
    • Try this new format until Toronto IETF90. The decide if we keep or move back to the previous format.
    • General support on the ML.
    • Comments on this call?
      • [Raghu] +1
  • [08.18] 6top-2-6top protocol
    • a node needs to access MIB of another node.
    • need to define the mechanism to support that.
    • negotiation interface and management interfaces are needed. Management interface will handle information flows between nodes.
    • Use of CoAP representation of YANG model to be transported as the payload of a Wrapper IE (new IE)
    • CoAP Format:
      • the 6top YANG model will be mapped into the URI and Payload of the CoAP message
      • Wrapper IE consists of:
        • Length, SubID, Content of the Wrapper IE
    • Question: what should be included in the payload of the Wrapper IE?
    • options:
      • Full CoAP msg, including URI-host and URI-port options
      • Full CoAP msg, without URI-host or URI-port options. Can the parser be self-contained?
    • [Thomas] What does it mean that the parser is self-contained?
    • [Qin] parser in the 6top layer vs general CoAP parser. No sure that the parser can parse the CoAP message without certain fields.
    • [Thomas] the parser would live inside 6top without involvement of other layers
    • [Xavi] from parser's point of view, no problem. Delta encoding, so no problem with parser. From point of view of parser will work.
    • [Xavi] If PCE configures something similar, could we reuse code?
    • [Thomas] the 2 parts should be independent
    • [Qin] Security check: are both URI-options necessary for the security check?
    • [Thomas] If they are needed can be added before parsing as they are well known values.
    • [Raghuram] multiple-parsers not needed/ we need to understand how we can play with URI/host options, they are flexible so we can use mac addresses in there if we want. Review CoAP draft for URI definition.
    • [Xavi] do we need the mac address? It is already in the mac header
    • [Raghuram] else you will be changing the parser a lot from a normal parser. It is a trade off, options to consider
    • [Xavi] I think the parser can live without an option.
    • [Thomas] host and port are optional to specify in case of multihoming, multiple virtual hosts (servers) behind a same IP address.
    • [Xavi] we can continue the discussion. If we use sheme 3, how do we do security and access check, eg only parent can write in children MIB.
    • [Qin] how to deal with security and access check. Example: Children cannot write parent MIB.
    • [Thomas] ACE deals with that.
    • [Thomas] that's a main point in ACE no everybody can hit a given resource but have an authorizing agent in the middle. Let us defer to ACE. The access check should be done before parsing.
    • [Thomas] does this make sense any sense at all? We borrow a parser from another layer.
    • [Raghuram] borrowing code is doable, right track
    • [Xavi] like the idea, need to find other examples in other protocols
    • [Pouria] URI-Port issue. Multiple end points in the same IP then URI-Port/host is used to differentiate them, this is an option. We need to worry about the source IP address as we are l2 and we do not have IP at that layer. So maybe we need to create something based on the MAC layer.
    • [Pouria] do we need OBSERVE in 6top to 6top communication?
    • [Thomas] We cannot decide yet. We need the whole CoAP implementation in case we want to support observe mechanism.
    • [Qin] should we have a weekly call to continue this discussion
    • [Thomas] +1
    • [Xavi] +1
    • [Raghuram] +1

      Action item: Qin to announce 6top-to-6top call on ML, possibly creating Doodle to agree on date, and as Pascal to set up webex.

  • [08.44] Lessons learnt minimal draft

    Moved to next meeting by lack of time.

  • [08.45] Terminology
    • [Pascal] Patrick read drafts in great details. Spotted lots of inconsistencies.
    • Introduce the concept of CDU Channel Distribution Usage. Global understanding. Notion of time - continuous == ASN, slotframe = repeat over an over. slot in a slotframe is the position within the slotframe.
    • [Dominique] "Channel offset distribution matrix", not "channel distribution matrix"
    • [Thomas] what is the difference between the slotframe and CDU matrix?
    • [Pascal] the CDU matrix only describes the characteristics of the network, is like a "picture"
    • [Maria Rita] the child that gets some cell allocated in the chunk will also be free to allocate them to its own children?
    • [Pascal] No, cells cannot be shared in the interference domain.

      Action item: Patrick to send pointers to the ML where the terminology is not clear.

  • [09.09] reminder next calls
  • AOB [1min]

    No other business raised.

  • [09.10] Meeting ends

Updated