Minutes interim-2021-roll-01: Fri 16:30
minutes-interim-2021-roll-01-202101291630-00

Agenda Roll-Interim Meeting -20210129
https://codimd.ietf.org/notes-ietf-interim-2021-roll-01-roll
https://ietf.webex.com/ietf/j.php?MTID=m34b599e4c5ac49d9e4bc6128a4cc26e7

Time: 9:30 am Eastern Time = 2:30 pm UTC = 3:30 pm CET

  Topic                                                 Duration          
  Presenter
- Introduction/WG-Status:            15 min         [Ines/Dominique]
- draft-hushe-roll-dodag-metric:     30 min         [Huimin She]
- dao-projection status and options  30 min         [Pascal]
- Open Floor                                             45 min                
 Everyone

# Attendees
- Ines Robles
- Georgios Papadopoulos
- Remous-Aris Koutsiamanis
- Dominique Barthel
- Michael Richardson
- Huimin She
- Li Zhao
- Rahul Jadhav
- Amaury Bruniaux
- Pascal Thubert
- Alvaro Retana
- add your name here

# Meeting notes
Times in UTC

## Intro/status
[14:31]
- Note well
- Note taking
- Agenda bashing: no comment
- Status of drafts: please provide comments on enrollment priority draft
-

## draft-hushe-roll-dodag-metric (Li Zhao)
[14:35]
New DODAG metric.
DODAG size is proposed as a new metric, to help nodes decide which DODAG to
join. Li Zhao Presents the draft on behalf of authors Explains the value of
DODAG si Dominique: two bytes for size is enough? Pascal: size in exponential
units. Rahul: what about rank? could it be used to discourage node from joining
a DODAG of which the root has a full routing table? Li: WiSun has...different
weight Pascal: can DODAG size be conveyed by Enrollment Priority? can we merge
this work into that darft? (to save admin work later on) Rahul: Priority can be
changed (increased) while passed down the DODAG, this one cannot. Not quite the
same. Aris: had a draft on balancing traffic at each node within the DODAG, by
announcing number of children: This is different. Pascal: this one is about
Root routing table size. If nodes are identical and have periodic traffic, also
relates to network load. Its the case for WiSun, with electrical meters. MCR:
6TiSCH did similar stuff in EBs (outside of encrypted messages). In WiSun, do
you assume a node can hear advertisments from both DODAGs? LiZhao: first PAN
advertisement is not encrypted. MCR: look at roll-enrolment-priority, it's not
a metric but could be useful. I don't think what you want is a metric in a DMC,
because does not need to be evaluated at each hop down the DODAG. Pascal: merge
this into enrollment-priority, and some info on overall load as well. Pascal:
seems we still need to decide if we need this info, but in any case it should
be merged into another document. Georgios agrees. Ines: please bring the
discussion on the Mailing List.

## DAO projection, status and options (Pascal)
[15:03]
Pascal reworked draft a lot.
Track is now a DAODAG inside a DAODAG.
Node at ingress of track does encapsulation, similar to Root doing
encapsulation as an ingress point into the DODAG. 6man was clear that SRH can
only be inserted by the source of the packet (for error reporting). MCR: do you
mean not 6550 is used recursively? Pascal: exactly Pascal: now UseOfRPLInfo
applies just the same way. -16 has several examples. You can find them in the
ML as well. Can see the VIA as a multihop TIO. MCR: originally, adding SM to
non-storing mode network Pscal: now, we have both. 6LoRH compression of SRH,
usually shared 6LoRH header. How does Root know the topology? non-storing mode
and sibling info option. Psacl: which siblings to advertise (especially if
many)? IPR on that, can share.

Profile 1
SM segments within source routed main DODAG. We shorten the SRH by using the
segments.

Profile 2
same as profile 1, but SR-VIO to install tracks.
Installed by the ingress. Sgnalling is simpler.

Profiles 3+ are across the DODAG, need sibling info

Profile 3
satefull VIO, similar to Profile 1.

Review of Huimin's comments
Todos

## TODO
- chairs to go through meeting recording and capture TODOs here
- bring discussions to ML