Clone wiki

meetings / 130524_webex

Minutes Webex 24 May 2013, 6TSCH group

Note: timestamps in PDT.

Taking notes (using Etherpad)

  1. Xavi Vilajosana
  2. Dominique Barthel
  3. Thomas Watteyne

Present (alphabetically)

  1. Alfredo Grieco
  2. Dominique Barthel
  3. Guillaume Gaillard
  4. Maria Rita Palattella
  5. Pascal Thubert
  6. Pouria Zand
  7. Raghuram Sudhaakar
  8. Tina Tsou
  9. Thomas Watteyne
  10. Tom Phinney
  11. Xavi Vilajosana
  12. Yoshihiro Ohba

Recording

  • This webex was not recorded.

Slides

Agenda

  • BoF recap [Thomas] [5min]
  • webinar announcements [Thomas,Xavi] [5min]
  • Centralized routing requirements draft [Pascal] [10min + 5min Q&A]
  • updated 6tsch draft [Alfredo] [5min]
  • updated 6tus draft [Xavi] [10min]
  • label switching discussion [Thomas] [15min]
  • logo challenge! [Pascal] [15min]

Minutes

  • [08.08] Meeting starts
  • Thomas shows slides and presents agenda
  • BoF recap [Thomas]
    • Description of the BoF as discussed last week. Will be shared in bitbucket repo.
    • Active draft list: 5 expected
    • Expected assistence 60. 1 session 2.5h
    • No special requests.
    • 143 members in the ML
  • webinar announcements [Thomas,Xavi]
  • BoF and IETF meeting practical info [Pascal]
  • Centralized routing requirements draft [Pascal]
    • 5 expected sections already identified.
    • [Thomas] What is the scope? do we want to modify PCEP to accommodate the different traffic flows we identified, or keep the original PCEP scope, just make it better applicable to LLNs?
    • [Pascal] This is just the rquirements draft, we don't need to make this decision this early. JP will help us to figure out if PCEP will fit to the requirements.
    • [Thomas] Could JP start a thread on the ML so we can discuss PCEP suitability?
    • [Pascal] Will ask.
    • [Thomas] Is our goal to be done with this draft by the BoF?
    • [Pascal] It should be started. Have Requirements and applicability.
    • [Pascal] We wil need to distribute the work (Maria Rita, Xavi, JP and others might be interested). JP is editor, but anybody can volunteer to participate.
  • updated 6tsch draft [Alfredo]
    • new version of the draft published yesterday
    • overview, problem statement and goals.
    • align terminology compliant to terminology v1 draft.
    • adding comments from Dominique.
    • 3 authors rather than one editor and 2 contributors
  • updated 6tus draft [Xavi]
    • small changes throughou
    • major change: execution of hard cell command
    • [Pascal] We might rename to "sublayer".
    • [Xavi] Will do.
    • [Thomas] All drafts hosted in bitbucket. Bitbucket has a ticketting system. Let's use it to track bugs.
  • label switching discussion [Thomas]
    • Discussion still open.
      • Each packet is associated to a label which is implicit
      • Each label is the position of the cell in the schedule.
    • Hop limit problem: TTL does not get decremented at each hop. Possible solutions.
      • As tunneling (dont' decrement)
      • explicit hop limit carried inline
      • use other info to determine the num of hops
      • [Pascal] hop limit is used to avoid loops. If the track is set up in a deterministic fashion, there are no loops. OK not to deal with this.
      • [Pascal] Label switching nice because can transport 6LoWPAN fragments.
      • [Thomas] If packet is routed by IP, a packet will have different hop limit than a tunneled packet. Without hop limit, we might not be able to do some optimizations.
      • [Pascal] the value is to avoid loops, as we don't have loops in tracks (if we can guarantee that), the ttl is not needed.
      • [Tom] Too early to make a decision, having the options is ok. Maybe we can have optional header.
      • [Thomas] We can conclude that there is no problem with tunneling so do not decrementing hop limit. It does not require an extra header.
      • [Alfredo] industrial apps: timestamp very useful. Since TSCH, all nodes are synchronized, lowering the field to 2.5 layer and then we can use the timestamp for hop count.
        • se timestamp to discover loops if it is higher than expected latency for the track.
        • this related to having the expected delivery time.
      • [Pascal] according to expected delivery time we can adapt priority of the packet.
      • [Thomas] you can drop packets if they are too late. Time sensitive delivery.
      • [Thomas] Alfredo, can start thread on ML?
      • [Alfredo] Will do.
    • Load Balancing problem: bundles can be formed by multiple in and multiple out. First cell will be used more than others. This can bias statistics.
      • Ideas:
        • round robin.
        • minimize latency and buffering
      • [Xavi] let it open to implementation but keep it as a point that 6tus addresses without defining the policy.
  • logo challenge! [Pascal]
    • selecting 2, 6,8 for a final contest.
    • vote on ML, then refine the final logo
  • [09.10] Meeting ends.

Updated