Date: Wed 2020-12-09
Slides: https://datatracker.ietf.org/meeting/interim-2020-detnet-03/session/detnet
Notes & BlueSheet https://codimd.ietf.org/notes-ietf-interim-2020-detnet-03-detnet
WebEx https://ietf.webex.com/ietf/j.php?MTID=mbee2c00f0fa9354e36f509cdc82372bd
Jabber: xmpp:detnet@jabber.ietf.org?join
Available post session:
Recording:
Presentation Information
Presenter: Chairs
Presenter: Xuesong Geng
Draft: https://tools.ietf.org/html/draft-malis-detnet-controller-plane-framework-05
Lou Berger: Comments on mgmt and OAM - since we now have OAM drafts, wouldn’t just a reference suffice? So we would want to put less about OAM in the controller plane draft rather than more.
Xuesong Geng: Agree, intent was just to have a reference to existing OAM doc. Do want to add managment aspects.
Lou Berger:Why would controller plane discuss management plane?
Xuesong Geng: Actually I am not sure of scope of management plane. Control plane is clearly configuration etc. e.g. by YANG. Want to discuss today - what is expected for this part?
Lou Berger: Agree that controller plane can be confusing since it needs to support both centralized and distributied control plane. Looking forward to new text to clarify.
Xuesong Geng: Note that RFC7426 about SDN architecture has control plane managment info, also there is info in RFC8655, so I can post some ideas based on that to WG list to get feedback from WG. We are now proposing adoption, but we understand that we need to clarify the relation of this draft to OAM. We propose to adopt first then work on management part. Would like to initiate weekly work session on this. Feedback welcome.
Lou Berger: Are there any concerns in the room about adopting this version? Personally would like to see planned changes done first, e.g. OAM clarification, but OK if WG decides to adopt first.
Any concerns for adoption? (We will do official poll for adoption on list).
No concerns expressed, so chairs will do adoption call, probably after new year (but possibly earlier).
Presenter: Dirk Trossen
Draft: https://tools.ietf.org/html/draft-trossen-detnet-control-signaling-00
Balasz Varga: Thanks for addressing my comments so far, I understand better now the focus. But the title of the draft is confusing - it sounds like it would address how to provide control plane signalling, but based on preso and comments, it is a specific scenario of DetNet nodes connected by TSN, and about how to use (and modify) RSVP to do this.
Dirk Trossen: Yes this is one use case. But yes, we are already expecting to rename the draft, and are currently looking at a new title for draft.
Balasz Varga: For the DetNet data plane we have split the info into individual drafts to address various aspects; maybe you will need more than one draft here to address the various scenarios.
Dirk Trossen: Agree, that’s why we propose to put use cases here, to set scope.
Balasz Varga: Regarding diagrams, I will propose some.
Janos Farkas: Question on RSVP DetNet protocol specification - do you want to define extensions to RSVP?
Dirk Trossen: Yes.
Janos Farkas: Then we need to discuss where in IETF to do that. Usually in IETF we do work in home working group, e.g. RSVP has its own WG, so we would need to discuss with them.
David Black: TSV WG in Transport area would be place for this but we don’t want it…
Lou Berger: Actually TEAS is responsible for RSVP now, both RSVP and RSVP-TE. So TEAS would be the right WG for this work. But for now we can keep all content in this document while the material is being developed. We can also address if the document needs to be split (and moved to TEAS), then before adoption we would take it there.
David Black: OK.
Lou Berger: The term “RSVP-DetNet” is confusing - I expected that it would be extensions to RSVP to support DetNet, as opposed to TSN. Please cover extensions for TSN separate from extensions for DetNet.
Dirk Trossen: Agree. Don’t have an exact plan for this now but will figure it out.
Janos Farkas: What signalling and associated RSVP extensions are needed to support DetNet? I can’t immediately see the next step - we are working on a control plane framework draft, which we are planning for adoption - are you thinking of working out RSVP-DetNet, i.e. the overall signalling for DetNet, or only TSN interworking?
Dirk Trossen: Need to discuss with co-authors, but expect to address both, since need to address both together, probably DetNet signalling first.
Presenter: Zongpeng Du, China Mobile
Draft: https://tools.ietf.org/html/draft-du-detnet-layer3-low-latency-01
Janos Farkas: FYI, regarding the first diagram, note that there is a section in DetNet Use Cases on Cellular Networks - this has always been in scope for DetNet.
Lou Berger: Are you looking at this as a standardized mechanism, or an informational doc?
Zongpeng Du: Solution is not limited, so informational is better.
Lou Berger: Good, that matches what we did with the bounded latency draft. But you didn’t say if you intend to implement a standard queueing mechanism or not - is this an implementation or a queueing mechanism problem? If the latter, it might be IEEE or TSV (David Black group). But in any case we are looking forward to the added content that you discuss.
Janos Farkas: It seems that the draft is discussing multiple problems to be solved, e.g. network scalability, time sync not being applicable to large networks, and microburst is yet another problem. Maybe it would be good to clarify which solution matches each problem; so far in the draft this is a bit confusing.
Zongpeng Du: We can add some additional explanation about these.
Janos Farkas: There have been a couple of queueing mechanisms on the table recently in IEEE 802.1, but not all are being standardized - e.g. the “Pater Noster” shaper - it has similar goals (to decrease state to be maintained) - maybe you can look at that - I can give you pointers if you are interested.
Zongpeng Du: Also we are doing some simulations of this, maybe we can discuss further offline.
Lou Berger: Any other topics before closing meeting?
Janos Farkas: FYI We have already scheduled regular YANG webex work session meetings and are continuing them in same time slot. Will have some break for holiday season, but that is already announced on the list.
Lou Berger: We would like to wrap up the YANG work soon, thanks for this work. Maybe start adoption for control plane doc before end of year? Can discuss.
Janos Farkas: Happy holidays and goodbye. Meeting adjourned.
Lou Berger, LabN
Pascal Thubert, Cisco
Ethan Grossman, Dolby
Janos Farkas, Ericsson
Balázs Varga, Ericsson
Xuesong Geng, Huawei
Dirk Trossen, Huawei
Yuji Tochio, Fujitsu
Don Fedyk, LabN
Yeoncheol Ryoo, ETRI
Andy Malis, Malis Consulting
David Black, Dell EMC
Deborah Brungard, AT&T
Fan Yang,
Franz-Josef Goetz, Siemens
Juergen Schmitt, Siemens
Stewart Bryant, Futurewei US
Zongpeng Du, China Mobile