Minutes IETF99: mpls
minutes-99-mpls-00

Welcome to Etherpad!

This pad text is synchronized as you type, so that everyone viewing this page
sees the same text. This allows you to collaborate seamlessly on documents!

Get involved with Etherpad at http://etherpad.org

1 . Agenda bashing, WG status reports    Presenter:  Chairs

Disclose IPR as soon as known.

George is not travelling this time.
Tarek flight cancelled and not here. Should arrive later today.

Jeff Hass: Could you please move BFD presentation earlier today instead of the
last. Loa: Moved to slot #4 (at the Mach's presentation)

Errata approved for RFC3031/5002.
3 new RFCs.
One new WG document.

Opportunist encryption: will expire. Need student.
Stewart: what is the issue?
Loa: EXP RFC. Couldn't find a student, had lost one. No technical issue.

2. draft-xu-mpls-unified-source-routing-instruction -  Presenter: Shaowen Ma

Unified source routing MPLS label stacking.
TE used as overlay.

Bruno: Orange: What document trying to specify can be summarised in one
sentencce. Node MPLS not capable, can use IP tunnel. Shaowen Ma: Use TE stack.
You can have the per hop behaviour. Bruno: Can use MPLS today. Shaowen Ma:
purpose of this draft is how to leverage existing capability. Informational
draft. Bruno: It is standards track document.

Shaowen Ma: We want to reduce the MPLS stack. No need to add MPLS labels.
Remember UDP port source value. Reuse the cached value. Difference from
exsiting forwarding capability. Bruno: MPLS and UDP is already specified.
Xiaohu: MPLS over UDP packet. No need to add reachble label. Loa: continue on
the list.

Bruno: relatiosnhip between -??- draft.
Stewart Bryant: still talking. Need architecture. Not sure how many drafts it
will take.

Bruno: As spring co-chair, why MPS WG chosen?
Loa: Relation with multiple working groups. TEAS/SPRING WG could be interested.
Bruno: This draft will have a better feedback in spring.

Ahmed Bashandi: What is new here. Running SR over IP tunnel?
Xiaohu: There is no need to have explicit tunnel. MPLS in UDP.
Loa: Take it to the list.

??:  Push label stack on the host. MSD is not an issue.
Shaowen Ma: Middle node is an issue pushing the stack.

3 draft-xu-mpls-service-chaining - Presenter: Xiaohu

No questions.

4 draft-ietf-mpls-bfd-directed - Presenter: Greg Mirsky, Carlos (Cisco)

First, Carlos Pignataro Presenting:

Stewart: Strict source routing from A to G? Due to ECMP path can be any
direction. Carlos: No strict paths.

Stewart: falsely detect a failure. Failure forward or reverse still a failure.
Carlos: Ability to select multiple path and correlate needed - there is
ambiguity.

Jeff Hass: Not required operationally.  (comment on the Slide before the
switching over to a different )

Jeff Haas: 5884 how to bootstrap a session - updates the 5884 procedure.

Stewart: Why do we want to modify and start a new one.
Jeff Haas: Using BFD. Doing unidirectional test. Failure on the return path can
falsely deletect the failure.  This does not always work. BFD is always
bidirectional.

Second, Greg Presenting:

In some cases create co-routed BFD session. Select more reliable reverse path.
Can use reverse LSP, TP like bidirectional LSP but does not have IP encap.

Carlos: 5884 can use LSP return path if bidirectional LSP.
Greg: providing a mechnaism to make it detereministic.

Loa: Do you support / agreement reached?
Carlos: Upto the WG. Not all questions answered.

Jeff Haas: 1. MPLS WG says useful prob to solve. 2. Issues in the document and
how to address if 1 is yes. 3. Minimal 5884 procedureal updates. Tackle in that
order.

5 draft-ietf-mpls-lsp-ping-lag-multipath - Presenter: Mach Chen

Loa: 3.5 years old. We have a 6 people.
Agree with Mach. Need much more WG document. Comments on the list.

6   draft-bryant-mpls-sfl-framework - Presenter:Stewart Bryant

Synonyous labels. Present all 4 drafts.

Loadbalancing issue:
Carlos: what are chances of that happening?
Stewart: check the deployment and see if that is an issue.
Carlos: How important is this limitation for this document?
Stewart: Strictly entropy label only
Carlos: ??
Stewart: need to announce to fulful this

George: In some older system, specific LC, egress and ingress LCs.
Stewart: ??
George: Yes, ELI include stack depth for load balancing.
Stewart: There are some routers that do that - use ELI and other things. Lot of
poeple not expecting this behaviour.

7 draft-sitaraman-mpls-rsvp-shared-labels - Presenter: Harish Sitaraman

Jeff Tantsura:  Whys stack labels?
Harish: Forwarding plane decoupled from data plane.  FIB size.
Jeff: motivation?
Harish: amount of FIB size.
Loa: Early allocation only WG document?
Deborah: Yes, has to be WG document, has to be stable document.
Loa: Ealy allocation premature. have to wait.
Pavan: Co-authos asking for adoption and then get early allocation code points.

8  draft-ietf-mpls-ldp-yang - Presenter: Xufeng

Loa: Yang dr email came late.
Loa: we are on track to update.

------------------------------------
Lou: New IPR rules.
Greg: How to decide the Evaluation and relevance.
Lou: WG /chair decides what to do with IPR. And evaluate the disclosures.
Text/paragraphs can be removed from the draft.
WG may not work on that technology.

Stewart: why present here?
Loa: have a document that needs this discussions.
Stewart: have a problem?
Loa: WG draft/LC. Got IPR. Have withdrawn the comments and draft changed. IPR
still holds. Have over disclosure in this case? Lou: Change / continue the
document. Lou: as a WG you can look at the document and say not going to
support it. Stewart: Difficult to learn the licensing term. Need court to
decide.
-------------------------------------

MPLS WG session II - Friday
1 - Kireeti Presents
Greg Mirsky: well known problem in IGP in split areas
Loa: for a WG document? there is no discussions on the mailing list?
Kireeti: did not know this is relevant to rings.. we talked to customers.. once
we make a deceision we will bring it to the list.. present in the update..
Kireeti: is the WG interested in this problem? Loa: I have not seen any
discussions on the mailing list Kireeti: So far, discussions on how to extend
RSVP for RMR happened in/with the WG

Questions:
    - none.

2. Greg Mirsky presents
a. draft-mirsky-spring-bfd
Questions: none
b. Point-to-Point MPLS LSP
Loa: send an email to get the WG's opinion on continuing this work in MPLS or
BFD WG c. draft-mirsky-mpls-p2mp-bfd Questions/comments:
    - Greg: I think this work more belongs to MPLS WG

3. Tarek Saad presnets
draft-ietf-mpls-static-yang
Greg: we need to think about synonymous flow labels
Lou: less of a concern about TTL, but TC seems suitable to have in the label
stack

4. Adrian Farrel presents:
draft-bryant-mpls-unified-ip-sr
Q: I'd like you to present to SPRING, as it is interesting to SPRING too. More
OAM part to be announced.. Adrian: we spoke to SPRING about this.. and we were
on the agenda this time.. on where it should be MPLS/SPRING WG? BRUNO: for
consistency anything we do for MPLS, can be done for SRV6. Adrian: less than
half of the work is in SPRING at the moment LOa: can we take this offline to
the list Stewart: hopefully, we can inherit a bunch of MPLS OAM, but it need to
be worked