Minutes IETF120: opsawg: Tue 16:30
minutes-120-opsawg-202407231630-02

What: Combined OpsAWG / OpsArea

When: 09:30 - 11:00 Tuesday Morning Session I, July 23, 2024

Where: Plaza B

OpsAWG Section

Administrivia - scribes, minutes, etc.
Joe / Henk
10 minutes

Warren: Anybody interested for serving as OPS Area director?

Agenda is fine, no comment.

Export of On-Path Delay in IPFIX
Benoît Claise
Draft:
https://datatracker.ietf.org/doc/draft-ietf-opsawg-ipfix-on-path-telemetry/

10 minutes

Joe: I agree, you guys have been responsive, we can kick off the last
call.

A Data Manifest for Contextualized Telemetry Data (9:45)
Jean Quilbeuf
Draft:
https://datatracker.ietf.org/doc/draft-ietf-opsawg-collected-data-manifest/

10 minutes

Benoit: let’s drop the normative dependency on full-include

Mohamed Boucadair: I would remove the dependency on
I-D.jouqui-netmod-yang-full-include but explain the expected behavior +
provide the fullback in slide 5 as an example. I don't think the draft
has to be prescriptive on that matter.

Jan: We need to collaborate indeed.

An Information Model for Packet Discard Reporting (Start 9:54)
John Evans
Draft:
https://datatracker.ietf.org/doc/draft-ietf-opsawg-discardmodel/
10 minutes

Rob: Why 48-bits counters?

John: Valid question, orginally was 64 then feedback from vendors was
it’s more that what’s needed.

Rob: in my experience it’s either 32 or 64 bits.

Diego: I don’t see many things that are required. The unusual nature of
this draft is interesting. A trend in YANG is to use several level of
abstractions to model the same thing.

Thomas Graf: module is very complete. Possible next step is how this
model can be transformed into a data model.

John:

Mahesh: RFC8343, model for interface, already defines a discard in the
YANG model. What does this draft add?

John: there are multiple causes of discards, that information is needed
to fix the issue.

Mahesh: We need to state that this model is part of a bigger existing
model for interfaces

John: next draft would explain how the data is exported and make that
link. Here we only focus on how to model the data.

Joe: we need guidance on how to define YANG modules

IPFIX Alternate Marking Information (10:06)
Giuseppe Fioccola
Draft:
https://datatracker.ietf.org/doc/draft-ietf-opsawg-ipfix-alt-mark/
5 minutes

Benoit: Flow key is not defined in IANA but depends on the use case.

Giuseppe: True, they can be promoted as flow keys, but they are not.

=== End Adopted Work Section ===

A YANG Model for Terminal Access Controller Access-Control System Plus
(TACACS+) over TLS 1.3 (10:12)
Mohamed Boucadair (Remote)
Draft:
https://datatracker.ietf.org/doc/draft-boucadair-opsawg-secure-tacacs-yang/

10 minutes

Rob: I would go for the reuse approach.

Pool about the adoption: raise hand, if you are in support of adoption
Yes: 14
No: 0
No opinion: 12

Mahesh: is this for the bis option?

Joe: it’s for adoption, we’ll see late if adopted

Henk: So no opposition, but significant number of no opinion. We need
more eyes on the draft.

Applying COSE Signatures for YANG Data Provenance (10:21)
Diego R. Lopez
Draft:
https://datatracker.ietf.org/doc/draft-lopez-opsawg-yang-provenance/
5 minutes

Mahesh: Adoption, what about NETMOD?
Joe: NETMOD chairs expressed interest in having the draft in NETMOD.
Maybe go to NETMOD after the implementation is done.

Mahesh: Good idea to use transaction idea. Need to collaborate with
Diego:

Per: Why do you use one signature only?
Diego: we though it was easier for us, we could

Kent (as NETMOD chair): more willingness that interest in NETMOD
Kent (as contributor): signature is independent of the encoding
Diego: we use base64 to encode the signature, it might be a little
longer

Joe: about adoption, we’ll find a home for this draft, there’s interest
in it

A YANG Data Model for Network Diagnosis by Scheduling Sequences of OAM
Tests (10:31)
Victor Lopez
Draft:
https://datatracker.ietf.org/doc/draft-contreras-opsawg-scheduling-oam-tests/

10 minutes

Greg: Very interesting work. The TWAMP standard does not allow control
of TWAMP-lite (given that it is a non-standard mode of TWAMP).
Moreover, you should look into the STAMP YANG module which allows
for scheduling of periodic measurement sessions.

Victor: Thanks, we’ll have a look, the goal for this model is to be
generic and encompass all OAM methods.

Joe (as a contributor): Why is the state machine read write?
Victor: Yes it’s mostly read to get the status, but you need to start
the test as well, I will think about it.

An Intent-Based Management Framework for Software-Defined Vehicles in
Intelligent Transportation Systems (10:40)
Jaehoon Paul Jeong
Draft:
https://datatracker.ietf.org/doc/html/draft-jeong-opsawg-intent-based-sdv-framework-02

10 minutes

Joe (as chair): I didn’t see any discussion on the list. I am not sure
we have expertise in the WG. As a contributor, looking at the slides
Jaehoon: WG should focuss on the network function part
Joe: Make sure to state that on the mailing list.

Mahesh: Break this work into several pieces and bring only SDN or the
management part to this or other WGs within OPS. Find another place for
the automotive specifics.

A YANG Data Model for Resource Performance Monitoring (10:50)
Mingshuang Jin
Draft:
https://datatracker.ietf.org/doc/draft-yu-ccamp-resource-pm-yang/
10 minutes

Rob: Is this related to the SAIN architecture?

Joe: Is there interest in working on this in OPSArea?
Poll: raise hand, if you have interest in working on this in OPSAWG
yes: 10
no: 0
no opinion: 18

Rob: if it is the same as the SAIN architecture, maybe we should point
out the differences.

Joe: nobody opposed, we need to see more traffic on the mailing list
before considering adoption.

Ops-Area Section

Administrivia - scribes, minutes, etc. (11:00)
Warren / Mahesh

Mahesh: IAB workshop, will be presented by Benoît
Mahesh: Green BoF tomorrow at 1 pm.

Warren: Again, if you want to run for OPS AD, please contact me.

IAB Workshop Redux on Management Requirements (11:02)
Benoit Claise
10 minutes

Warren: I am assuming you will send reminders?
Benoit: Yes, to multiple mailing lists

Joe (as a contributor): you have to submit a paper to be invited, can
you go even if you don’t have a paper?
Benoit: It’s not going to work for operators to submit a paper. There is
theory and there is practice

Joe: why are dates TBD?
Benoit: just need IAB to press the button

Laurent (as NMRG co-chair):
Benoit: we didn’t validate the call yet, because the distinction between
IRTF and IETF is not clear

Laurent: we need to focus on approaches that worked or not in the past
and capture new approaches that might impact the network management.

Diego: One joint contribution from all people in Telefonica or multiple
ones?

Benoit: yes :-)

Rüdiger: I would like to reinforce Benoit point about the rules being
relaxed. Inviting the right people is hard. Make it clear that you are
liberal and encouraging operators to come without burden, especially
true for participants coming out of external orgs. I could not have
survived the previous workshop if I had had to write papers.

Mahesh: distinction between NMRG and what this workshop is trying to do.
"Why not AI" did come up often. The workshop is about getting
requirements from the operators. We are not trying to go with an opinion
of what we believe the networks should look like.

Benoit: Maybe you want to check whether the description of the workshop
is clear enough.

Mahesh: In good shape but maybe need to clarify the difference between
NMRG and this workshop:

Oscar: These external organizations. Plan to have formal invite/input
from other SDOs (TMF)?
Benoit: (not necessarily) Not clear that we want SDOs communications as
they would want to push for their agendas rather than get feedback from
operators. Slightly concern if we have SDO formal communication

Dan Voyer: Hard to get requirements from operators, as they are too many
works in parallel and it's messy. The fail fast appproach... NMOP should
be centralizing that role, otherwise it’s hard for operators to go to
the right place and the message might get lost.

Benoit: I agree on the step-by-step an the connection with NMOP. Maybe
there are bigger things than what we are tackling in NMOP.

Mahesh: As Benoit noted, we see a decline in the number of operators
coming to IETF, so outreach is important.

Dan Voyer: I think this is the way to go

Open Mic (11:26)
(as required)

Benoit: you were mentionning the OPS directorate. It’s scary to tell
people what to do. You provide feedback to the AD, you are not liable to
make it happen.

Warren: That’s a good point, these reviews help the ADs identify
potential areas of concern in the drafts and can then work to push those
issues to a resolution or a mutual understanding.