Master at: https://notes.ietf.org/notes-ietf-119-anima
IETF119 is a hybrid IETF meeting.
Please familiarize yourself with the new&changed procedures for
participating in-person or remote:
https://www.ietf.org/how/meetings/preparation/
ANIMA will have one 1 hour session,
chaired by Sheng Jiang and Toerless Eckert
17:30 - 18:30 (local time AEST, 07:30 - 08:30 UTC) Monday Session 4,
Room M4
Number of participants on meetecho: 28 (at start). In the room: 13 (at
start).
Note taker: INSERT YOUR NAME HERE
Time: 5 minutes
Presenter: Sheng Jiang (remote), Toerless Eckert (local)
Work since IETF119, Administrative, Welcome to new AD
Time: 10 min
jws-voucher finally not including new mediaman type.
brski-cloud: own and michael have about 17 open issues that need to be
closed before it can be brought up back to AD review.
Rob: jws-voucher can be kept in WG until the dependencies move forward.
Esko: constrained join proxy, Esko not yet author, but could become one.
Bug in slide: brski-discovery 2024, fix TODO
Draft: draft-ietf-anima-brski-ae-10 (was -09 at IETF118)
Presenter: David von Oheimb (remote)
Time: 5 min
Apologies from Rob for not getting to it before change of AD.
Draft: draft-ietf-anima-brski-prm-12 (was at IETF118)
Presenter: Steffen Fries (remote)
Michael Richardson: Interesting new use-case, transport over USB stick
(from hackathon). Unclear whether it would mean any actual change to the
document. Hopefully not. If it can work, it could be a big use-case.
David: Works over any transport including sneaker-net.
Michael: I have to convince myself that is the case and it doesn't need
any changes.
Presenter: Michael Richardson (local) / Toerless Eckert (local)
Time: 10 minutes
Toerless: Main issue is that there are no tool-chains to validate
YAND->CBOR/JSON.
Should we change to CDDL specification (without hurting backward
compatibility of course).
Esko: Only/main other draft using YANG for specification similar is
"CORECONF" https://datatracker.ietf.org/doc/html/draft-ietf-core-comi-17
a CoAP-with-CBOR version of NETCONF/RESTCONF
Toerless Explaining (without slides) that we still have not managed to
have compile toolchain for the YANG. In addition, CDDL is a theoretical
option as we only need CBOR/JSON compilations. So we may want to explore
what participants feel about change to that options.
Kent Watson seemingly proposed a way to get tool chain (Yang-Lint) to
compile the YANG model. Michael wants to try that out.
The other thing, there is a proposal for a full include mechanism to
YANG. But this would probably add a lot of time. Perhaps now is not the
time to make this change.
rfc8366(bis) is not q protocol. Its a digital artefact with description
of some semantics and some operations such as signing. But other
RFC/Draft describe how they use it (RFC8995 and other anima/netconf
drafts/RFCs)
Conclusion (Rob/Kent): Do not split it up, keep it as a single doc as it
is right now.
Counter-example from Michael: rfc9417/9418 where spec/yang where split
up. But unclear if/how this would apply to us
Michae: Thinks there will be anothrer rev in maybe 2 years, so maybe
structural improvements like what Shepherd (Alex) proposed could be done
then so we do not prolong the current work.
Mahesh: I agree that splitting the documents would be a good idea, but
we need to get IETF consensus. There is no East-West protocol that can
use YANG. Please can you clarify what you mean by that.
Toerless: We have lots of North-South protocols but we have no East-West
protocols. This might be harder for tooling.
Rob: Certainly true that this use is not the primary intent of YANG, so
unclear what to do going forwarding. Maybe we'd need some IETF ASN.1 new
language.
Carsten Borman: many modelling languages, e.g.: W3C schema. For very
simple artefacts like voucher, CDDL should work very well. When it
becomes more complicated, more tooling may be necessary.
Vocher originally was part of Zero Touch Bootstrap draft and that was
north/south, so when it was extracted, it became east/west within BRSKI.
Suggest a PiYang method to compile .
Michael Richardson: Did try some of the possible YANG fixups to make
YANG extensible as we wanted it, but it didn't work in two dimensions,
thats why we pulled partial YANG pieces from multiple drafts into common
rfc8366. But the experience was the YANG did not help any coding.
Switching of YANG to CDDL would not incur any development cost, because
no YANG was useable in development (yet)!
Some CDDL tools exist to create header files. Unlikely quickly revise
existing implementation coding process.
Primarily we need to be 100% backward compatible in the serialization.
That applies to all YANG changes.
Michael thinks that with the way Kent proposes, he can now verify that
the serialization of RFC8995 will be exactly the same (for existing
elements) like new rfc8366bis and equally whether Alex' proposals would
not change serialization.
But already spend over 1 years mucking around with YANG so we should
call it quits on changing/improving mayor spec parts for architectural
reasons.
Draft: No draft yet associated.
Presenter: Sheng Jiang
Time: 10 minutes
This presentation was on agenda for IETF118 but we ran out of time.
Michael: WOuld like to see more details.
Rob: This likely requires re-charter if this was to be adopted.
Rob: Likely also too big for going directly for a non-chartered WG.
Brian Carpenter: running GRASP over TCP works fine. Still needs to be
convinced of the benefit switching away from it. Aka: have doubts of
need for this.
Bing: I believe that we have a requirement to decouple as you suggest in
the slides. When doing some internal work, we found many cases that
don't support IPv6 or even L3 routing. I would like to remove any
dependency to make GRASP more flexible.
Carsten: Although everything you say, points you towards COAP, which is
really lightweight.
Toerless: Even DTLS has retransmissions.
Carsten
Toerless: Am I wrong.
Michael: Only the ,,,
Draft: No draft yet associated.
Presenter: Toerless Eckert
Time: 10 minutes
Time: (1 minutes)
Discuss options for CBOR style discovery for BRSKI-discovery with
CBOR experts
TODO: Chair to ask WG to add Esko as co-author to constrained join
proxy. Brian thinks no permission is required.
TODO: Provide more productive feeddback for a new version for
network-service-auto-deployment. Maybe have a call with authors.