Minutes for LIME at IETF-96
minutes-96-lime-1

IETF 96
IETF 96, Berlin, Germany, July 17-22, 2016

LIME
Layer Independent OAM Management in the Multi-Layer Environment

Thursday Morning session I
10:00-12:30
Room:  Tiergarten

[Meeting is starting. Carlos is chairing the meeting, Ron is not here this
time.] Note well applies.

Agenda:

- Administrative - Chairs

  . https://datatracker.ietf.org/wg/lime/charter/

[Carlos giving an overview of WG status.]
WG has a wiki that contains information on applicability and usability of LIME
models. Please take a look.

- Data Models:

  . draft-ietf-lime-yang-oam-model-07
    ÒGeneric YANG Data Model for Connection Oriented Operations,
    Administration, and Maintenance(OAM) protocolsÓ
    https://datatracker.ietf.org/doc/draft-ietf-lime-yang-oam-model/

    Review WGLC Comments and pending issues.

[presentation]
Michael Wang presenting.

Multiple iterations of changes based on 2 interim meetings and multiple private
discussions.

[discussion]

Greg Mirsky/Ericsson: Why do we have different grouping for MEG-ID and MA-ID?
MEG-ID is a different type of MA-ID according to Y.1731. Michael: You can
choose whether to follow MPLS-TP terminology and define name format according
to MPLS-TP, but if you choose to follow TRILL format, you can set it to TRILL's
type. Greg: If you want to keep MPLS-TP as a separate technology, there is
already a draft for MPLS-TP OAM, and we should augment it by that model and not
redo it in this WG. Greg: Please take a look at that document and see how it
could be augmented. There is no need to replicate MPLS-TP OAM constructs in
LIME. Carlos: Are you saying that MPLS-TP draft and YANG model is referenced by
using these? Greg: No, I suggest the opposite - this work references MPLS-TP
work. Why is there a need to reference MPLS-TP specific construct here? Carlos:
LIME approach is that various consumers can use this model and not the
technology specific ones directly. Greg: It uses MPLS-TP specific constructs
rather than opaque identifiers, it provides constructs that are technology
specific rather that agnostic. Carlos: MPLS-TP needs to augment these
non-technology specific constructs and not the other way around. Greg: This
work looks that this is technology specific. Carlos: please work to remove
technology specifics. Reshad Rahman: Greg - I think your comment also applies
to other technology specific areas, not only to MPLS-TP? Benoit Claise: I
wonder how would you work with the topology? We have several topology drafts in
I2RS. In the topology draft there is a notion of node id. How do we relate the
node id in the topology to what is being discussed in OAM model? Michael: Here
we use connection-oriented technology approach that has such mapping. For
connectionless we need to work on finding such a mechanism. Carlos: There is
one action regarding the addressing - do an editing session with YANG doctors,
that would be really useful. Carlos: The comments related to addresses, to
topology - we need to spin a new version.

  . draft-kumar-lime-yang-connectionless-oam-03
    ÒGeneric YANG Data Model for Connection Less Operations,
    Administration, and Maintenance(OAM) protocolsÓ
    https://datatracker.ietf.org/doc/draft-kumar-lime-yang-connectionless-oam/

    Updates targeting adoption.

[presentation]
Qin Wu presenting.

Many changes and updates based on two interim meetings and individual reviews.

[discussion]

Greg Mirsky/Ericsson: If authors would like performance monitoring as a
separate model, then I would suggest it to be characterized as fault
management. It also does not need to mention OWAMP and TWAMP. Qin: We want to
have a more generic model and not certain whether we should change a name.
Greg: I would encourage to separate the models. It would make easier to manage
them. Otherwise we will have to talk about technology specific PM models too.
Qin: We need to have a generic name. Greg: It is a choice by authors, I cannot
force anything, just suggesting. Carlos: Can you discuss the relationship of
the document and other technology specific ones? BFD relationship in
particular? Qin: We clarified in this document some details on BFD interaction.
We will update the wiki applicability page too. Carlos: BFD is meeting
tomorrow. Reshad and Greg, what is the status of that document within BFD WG?
Reshad: There have been changes in routing model regarding instances. The
remaining part is the technologies that BFD depends on. You do not want to
redefine the way MPLS tunnels are configured, we should be reusing, augmenting,
and not redoing it from beginning. Greg: BFD model covers single and multihop
and LSP. Micro-BFD is there too. Work on LSP ping YANG model will be
reactivated. It will be discussed in PCE session. Carlos: It was a bit stale
previously. Greg: Based on request from MPLS WG chairs the work will be
restarted. Carlos with WG hat on: Time is pressing to find a real time editing
session on these more intricate issues. Greg: We should make effort to make it.
Interim meetings were very successful. We could schedule some interim sessions.
Frank Brockners: The document is good to adopt. Can you split the document into
methods to retrieve the data from the model itself, this would allow for other
data retrieval mechanisms and not to overload the RPCs. Carlos: It is a WG
decision. There are advantages and disadvantages of such approach. We need to
agree on the right granularity. This might be a question to YANG doctors.
Benoit: Are those two separate models at the moment? Qin: Two models in the
same draft. Benoit: Maybe it is a good step to separate it then. You may want
to update the model only. From the YANG perspective we could do it. No strong
opinion though. Carlos: Please take to the list, summarize the pros and cons of
having the same model, and discuss there. Benoit: How do you connect with
topology? Qin: We augment the network topology model. Benoit: Is that a model
from I2RS? Qin: Yes. For connectionless we have more restrictions to model the
topology. Deepak via jabber: As an author I am happy to break the draft into
two, we will wait for the decision from the WG.

Carlos: In the context of the document, we have been working for a while on
this document. This is still an individual document, we have marked as a
candidate for adoption. Before taking that question to the list, does anyone
have any concerns with it? This is within our charter. Are there any concerns
for this document to be a base for LIME connection less model? Greg Mirsky: The
number of times we talk about relationship between generic and specific models
is a very important question. Once we solve it we will have a very good basis
for the future work. Carlos: Could you clarify what do you mean with solve?
Greg: As a coauthor of MPLS-TP OAM document, I still see there are things that
there are things that we need to address - namespaces in particular. I havenÕt
looked at the latest version of applicability document though. Will look at it
and discuss.

- Discussion - All

- Next Steps - Chairs

Carlos: Last topic that we have today is on the relationship on the common
model and the technology specific ones. Some technologies are more advanced -
BFD, TRILL, MPLS. Some are not even talking about it yet. Carlos: Wiki contains
that information. You can also edit it. Benoit: TWAMP is IP only, right? So
that means connectionless? This is an open document. Benoit: Carlos, you were
asking on sending the document to the YANG doctors for review. What is your
conclusion? Carlos: We need to solve the generalization of the addresses first.
This is the only major aspect with the documents. After that I need to talk to
Ron, and we could send it to YANG doctors, and then issue WG LC on the
documents.

Carlos: Meeting ended.