[Bier] BIER WG Minutes, IETF 95

[Greg Shepherd <[e-mail]>]

Thanks to Tony for the excellent minutes. Please read and reply with
corrections. They will be submitted soon.

Thanks,
Greg

---minutes---

BIER Meeting Minutes

Exec Summary: OAM, Yang towards adoption, core drafts moving well towards
LC, security still missing, discussion whether it should moved into
standards track



   - Note Well
   - Greg: Yang: only points of contention and progress to be reported
      - towards last call, last point of contention scope of flooding of
      BIER info
      - no questions
   - Eckert: Biert-TE:
      - new co-authors
      - FRR: modifying bit mask as local decision and rerouting the packet
         - Mirsky: how do you differentiate between link & node failure ?
         can’t, like RSVP, probably either/or
      - subdomains
         - concerns about applying term of SI to BIER-TE
         - new terminology will be introduced and mappings analyzed
      - suggestion of introduction of slices (single bit with its own
      label)
      - Andrew/ALU: Is that of any value ? don’t make it WG item
      - OAM: we have OAM & that runs over resilient layer and this is
      violating multi-layer network OAM recommendation
      - Diego ? : in physical topologies possibly valuable ?
      - chairs: please take out to list as “protection for BIER” discussion
      - chairs: split TE from the “protection” discussion and what do we
      adopt
   - Mirsky: Performance Measurement with marking in BIER
      - double marking method more advanced & preferable
      - call for WG adoption
      - rough consensus of readers to adopt
      - call will be taken to the list
      - chairs: delayed question: this does NOT prohibit use of OAM bits in
      different manner ?
         - unclear: discussion to be taken to the list whether we can
         imagine other uses of those OAM bits ?
      - Mirsky: BFD in BIER
      - part of larger piece of work from the “overlay OAM design team”
      - submitted to RTG WG
   - Sandy?: BIER Ethernet Encapsulation
      - call for adoption:
         - Chairs/Eckert: where is the discussion ? what are the use cases
         ? who will do the work
         - Alia: where are the use-cases, we also have a Wiki ?
         - Andrew: generate more interest
         - chairs: we can’t just let it sit or worse, standardize without
         review.
         - Greg: IEEE is not expensive ;-)  Here’s a chance to get a 2.5
         multicast solution
         - Alia: talk to Pat Tayler
         - Chair: comments been given on list about MAC addresses. This
         NEEDS review
         - Eckerd: Wiki too light ?
         - Alia: yes, f2f let’s figure out how we work together. Form the
         discussion anyway you want
         - David: Where are the DC people with use cases so the claims can
         be unsubstantiated.
         - Partha/Bloomberg: Sounds interesting
         - David/Mellanox: BML MUST be set on all encapsulations
         - chair: needs more input/discussion/collaboration before we call
         adoption
      - Sandy: BIER Use Case in NVO3
      - Greg: is BIER overlay coordinated with NVO overlay ? BIER is a
      transport for NVO
   - Ran: Yang for BIER
      - chair: what’s the FRR in BIER ?
      - chair: great it’s validated, that’s real progress
      - Greg: ?
      - chair: should we call for adoption ? solid consensus for adoption.
      will be taken to the list
      - chair: FRR needs removed until we have consensus/WG items
   - Ran: BGP-LS for BIER
      - has been discussed with Peter, aligns with OSPF
      - weak consensus to call for adoption with no opposition. Will be
      taken to the list
   - Linda?/Path Autogeneration:
      - Chairs: read/comment
   - Linda?/BIER TE Yang Model:
      - Chair: Do not explain the model, highlight progress and issues
      - Greg: why not augment the base model ? BIER-TE fwd is very
      different ? But Yang is _configuration_ and _state_, fwd data plane
      paradigm is orthogonal to that. Question there are sufficient differences
      between the models here
      - Eckert/Cisco: ?
      - chair: get a doctor ? no. get a doctor.
   - Chair: Should we aim for Standards Track ? What’s the Criteria ?
      - chair: we lack security
      - Alia: I need to look @ problem statement to address it further.
         - Documents.
         - being EXPERIMENTAL helped focus and speed
         - implementation status added ? experience how it’s used ?
         applicability ?
      - Robert Raszuk: we can’t get silicon vendors to implement because
      experimental (as Bloomberg input) so it’s chicken/egg problem.
      - Alia: but then get me more status/comments on list. It will touch
      _everything_ in the hourglass once it becomes PS. So consider this.
      - Robert: we’re looking BIER in overlay so we can even software.
      - Alia: so add use-case to document



Vendor discussion:

   - MPLS draft should strongly MUST that the BML is set in the packet