Minutes for BIER at IETF-96
minutes-96-bier-1

BIER WG Minutes
IETF 96 Berlin
Monday, July 18, 2016

== Chairs ==
Discussion about Future encapsulation is from, quote, one more encap, in the
charter. Mpls ok, so discuss here which one to be. Documents are still in the
experimental track, demands to go for standard track. Need at least live
deployment to do so. Alia : wglc and review as a first serious conversation to
go forth GregS: how to handle doc publication Alia: leave in the queue without
requesting publication. Allow row modify the status on the fly

== BGP-LS ==
No comments. Raise of hands show support for adoption. Validation to go on the
mailing list

== DR/DF Election ==
Slide 6 : ".G" in the formula to be clarified later
Discussion on inconsistency periods, shall we go for a sticky behavior or not ?
Ice answer: both are possible
GregS: reason for the doc is not only bier, but people are asking for PIM-free
networks, so it was ask to gather and discuss requirements in bier. Can be
troublesome is some way, at least a review in other wg such as ospf or isis is
needed to avoid to break the protocols. Furthermore as we don't want in bier to
solve a global problem but to fit bier requirements, so maybe go for
information and address the general problem in IGMP. PIM chair: people in PIM
may be happy with that, just potential discussion on the hashing function. Ice
answer : don't forget the draft also addresses the DF function, not just DA.

== FRR extension ==
No questions.
GregS: Expectation to ask for adoption? - Yes
GregS: feedback is welcome and really needed

== Discussion on encapsulation exploration ==
1st question: what can't be addressed with the current architecture?

Tony: New Encapsulation helps how? Bier is 2.5
Pierre Pfister: don't see bier as 2.5 but as layer-3. Really like BIER but sad
not to be able to use natively in IPv6, need for another technology, and
between bier with MPLS and crappy-old multicast I choose the latter because
MPLS is too complicated Ice+Andrew Dolganow: MPLS use needs clarification. No
MPLS control-plane machinery here, just a reuse of bits

X.HU : need of more analysis to choose the right ones
GregS: wg is experimental, not standard track. It is not about the multiple
required encapsulation but just one to choose.

PierreP: choice of IPv6 in spring was for good reason.
GregCauchie: Simplicity is sought by everyone but simplicity is from where you
stand, for ISP-folks it is usually MPLS and for datacenter folks it is more
IPv6... so maybe there is no answer to "where MPLS breaks" as MPLS encoding can
be pushed anywhere, but rather this is a question of "which one is the more
natural candidate", and from spring discussions + bier history that followed
segment-routing for multicast, maybe at the end IPv6 is the more natural
candidate for a second encap.

Alia: beware of IPv6 use, may not work today, so charter says one more encap
but need for the group to achieve the charter and not be too distracted.

Tony: after all the discussion it seems second encap will be a fork lift.

Many body count to participate in a design team, will ask officially in the
list to throw things around. Interim WG meeting for design team to be announced
on the list - target is late Sept.

Adjourn