DetNet Minutes for IETF 115 (London and Online) ============================================== WG ICS: https://datatracker.ietf.org/meeting/upcoming.ics?filters=detnet Datatracker: https://datatracker.ietf.org/group/detnet/about/ # Monday, November 7, 2022 - Session II 13:00 - 15:00 (GMT/UTC) Richmond 4 Materials: https://datatracker.ietf.org/meeting/115/session/detnet Note taking: https://notes.ietf.org/notes-ietf-115-detnet#both Meetecho: https://meetings.conf.meetecho.com/ietf115/?group=detnet&short=detnet&item=1 Onsite tool: https://meetings.conf.meetecho.com/onsite115/?group=detnet&short=detnet&item=1 Audio stream: https://mp3.conf.meetecho.com/ietf115/detnet/1.m3u Jabber Logs: http://jabber.ietf.org/logs/detnet Session ICS: https://datatracker.ietf.org/meeting/115/session/29840.ics YouTube: https://www.youtube.com/watch?v=FxWBvlYnU1M Session Recording: https://play.conf.meetecho.com/Playout/?session=IETF115-DETNET-20221107-1300 ## Joint Session with PALS and MPLS on Wednesday, November 9, 2022 - Session I 9:30 - 11:30 (GMT/UTC) See Materials: https://datatracker.ietf.org/meeting/115/session/pals # Slot Start Time (UTC) Duration Information ## #1 13:00 10 min Title: Intro, WG Status, Draft Status ### Presenter: Chairs ## #2 13:10 10 min Title: LS on 3GPP 5G System acting as a Detnet node ### Presenter: Laurent Thiebaut ### Draft: https://datatracker.ietf.org/liaison/1800/ (start: 13:06) Xuesong Geng: As author of YANG model, I can try to give responses to questions. ... for E2E node delay attribution is (?) - (defer note to written response?) Lou Berger: We need to provide a written response to the liason. Janos Farkas: We will start with a response draft in working group list, if interested please write to the list. ## #3 13:20 10 min Title: Update on OAM Documents ### Presenter: Greg Mirsky ### Draft: https://datatracker.ietf.org/doc/html/draft-ietf-detnet-mpls-oam-08 Greg Mirsky: waiting on responses to comments raised in WG Last Call ### Draft: https://datatracker.ietf.org/doc/html/draft-ietf-detnet-ip-oam-05 (start at 13:18) Lou Berger: We have had discussions on how to do association with IP flows with OAM, for DetNet flows. UDP is one way, but others are allowed. You highlight use of DetNet DSCP, but it is allowed but there is no dedicated DetNet DSCP. So you need to make sure this draft is aligned with DetNet Data Plane doc, which allows use any of various flow marking methods. Greg Mirsky: To do fate sharing of OAM packets to match monitored flow, then the same marking must be applied to OAM packets. Lou Berger: What if network is not using DSCP for marking? Is that case considered/allowed per this draft? Greg Mirsky: Yes, the same applies to OAM - if the network is not using DSCP for the monitored flow, then DSCP doesn't matter for OAM packets. Just be clear about it - DSCP not the only methodology. Needs additional clarifying QoS policy of flow being montired, same must apply to OAM. Did you get Pascal's comments? Will follow up. Check minutes for notes and follow up on list. ## #4 13:30 5 min Title: Deterministic Networking (DetNet) Controller Plane Framework ### Presenter: Geng Xuesong ### Draft: https://datatracker.ietf.org/doc/html/draft-ietf-detnet-controller-plane-framework-02 (start time: 13:28) Carlos Bernardos: Will you be adding to this document more about the relation of DetNet controller plane to RAW? Xuesong Geng: This is just for requirements. Lou Berger: Expect it is reasonable to create a RAW-specific document. But if that text is small, could incorporate it into this doc. First step is to idenify RAW specific parts, then RAW and DetNet can decide whether to integrate into this draft or create new draft. Lou Berger: Reminder at the last meeting we discussed integrating Multi-domain text from Carlos Bernardos, we want to see that added before WG LC. ## #5 13:35 5 min Title: Deterministic Networking (DetNet): DetNet PREOF via MPLS over UDP/IP ### Presenter: Balázs Varga ### Draft: https://datatracker.ietf.org/doc/html/draft-ietf-detnet-mpls-over-ip-preof-01 (start at 13:35) Lou Berger: We started poll in the room: Please join queue if you have comments or objections now. (Raised hand means object) One hand raised - discuss now? --> No response 24 hands not raised, 25 total. ## #6 13:40 5 min Title: Deterministic Networking (DetNet): Packet Ordering Function ### Presenter: Balázs Varga ### Draft: https://datatracker.ietf.org/doc/html/draft-ietf-detnet-pof-00 (start at 13:39) Lou Berger: Starting live poll, same question - any objections? Raise hand if object, else not. Please only respond if you have read the doc. Result: No objections, but few have read draft. Ehsan Mohammadpour: We have done studies on reordering and POF, including work with additional parameters which are not included in this document. Balazs Varga: We have listed only mandatory parameters, for example buffer size can be calculated from flow characteristics, but some such parameters are important for implementation so could be added. Toerless Eckert: Have you done any validation of the mechanisms discussed in the draft? Balazs Varga: Yes, I know of some private validation activity ## #7 13:45 10 min Title: Requirements of large-scale deterministic network ### Presenter: Peng Liu ### Draft: https://datatracker.ietf.org/doc/html/draft-liu-detnet-large-scale-requirements-05 (start at 1346) Lou Berger: At last meeting (or other recent) we heard about the NEXT (?) draft, once this doc is adopted by WG, will you incorporate that NEXT text? Peng Liu: Yes we will. Lou Berger: Some of the items described in this document could be generic implementations for DetNet, not specific to large scale DetNet. It is mportant to distinguish in the document which are unique to large scale. Peng Liu: Agree. ## #8 14:55 9 min Title: DetNet Data Plane Enhancements for Large-Scale Deterministic Networks ### Presenter: Quan Xiong ### Draft: https://datatracker.ietf.org/doc/html/draft-xiong-detnet-large-scale-enhancements-01 (start at 13:55) Xuesong Geng: If there are any new requirements for controller plane that arise from large scale DetNet networks, please mention in the list and we can discuss further to include them. Also there are 2 docs on requirements for large scale DetNet - are the large scale req'ts the same as DetNet requirements? I would like further clarification. Lou Berger: This draft is specific to large-scale-driven enhancements to DetNet - It discusses clearly 2 items that are discussed in another draft, will bring in that text after adoption. A lot of this text will be moved to (or addressed in) other drafts, also controller plane draft, but there are useful pieeces of this draft even after breaking out those parts. Xuesong Geng: But are they in general the same req'ts? Lou Berger: Not making a generalization, but regarding specifically this draft, these items are intended to be specific to large scale. Daniel Huang: These are enhancements to DetNet architecture. I am aware of DetNet controller plane docs, but the massive flow issues and queue mechanisms that are needed for large scale are currently absent in current DetNet architecture docs. So these are new features of architecture for DetNet. Peng Liu: Some comments have been merged to other documents, but some will remain. ## #9 14:04 9 min Title: Asynchronous Deterministic Networking Framework for Large-Scale Networks ### Presenter: Jinoo Joung ### Draft: https://datatracker.ietf.org/doc/html/draft-joung-detnet-asynch-detnet-framework-01 (start at 14:05) Greg Mirsky: You say this mechanism provides guarantees of latency and jitter? Jinoo Joung: Just latency. Greg Mirsky: But is it a 100% guarantee to arrive within given credit? Will not work unless overprovisioned? Jinoo Joung: Actually there is provision for latency guarantee - total arrival rate can't exceed capacity of path. Greg Mirsky: Does each transit node maintain shaping amd policy? Jinoo Joung: No shaping is necessary here. Greg Mirsky: From my empirical experience, statistical multiplexing will affect the arrival expectations, so this technique may improve ontime delivery, but I'm skeptical that it will actually guarantee it. Jinoo Joung: It is known for 20 years as solution. When number of flows are many thousands, is difficult for very low latency values. Lin Han: I don't see if we need a reservation? Jinoo Joung: Yes, reservation is necessary. Lin Han: Then the reservation must specificy the required latency. Jinoo Joung: If reservation is properly done, latency is guaranteed. Lin Han: But if latency is very small, it is difficult to achieve. Jinoo Joung: I showed the equation. Lin Han: But value has to be quickly estimated, can't be a complex calculation in a real time system. Lou Berger: Must defer to list, out of time. A lot of interest in queueing mechanisms, maybe need interim to allow more time for technical discussion. ## #10 14:13 9 min Title: DetNet Data Plane - Tagged Cyclic Queuing and Forwarding (TCQF) for bounded latency with low jitter in large scale DetNets ### Presenter: Toerless Eckert ### Draft: https://datatracker.ietf.org/doc/html/draft-eckert-detnet-tcqf-00 Lou Berger: Do you care what the setup mechanism is? You just made a strong statement about this that I didn't get from your draft. Toerless Eckert: No. Well now I'm not sure, maybe can discuss offline. Greg Mirsky: In second slide you list multipath controlling technologies - is this intended to be an exhaustive list? Toerless Eckert: I don't know a stateless multipath. Maybe some other tricks to get there with BIER and SRv6 Greg Mirsky: In the same way, SRv6 can use explicit path routing. If you think this list is exhaustive, you did not mention BIER, just BIER-TE. Toerless Eckert: Possible to list all nodes, like in SRv6 (?) Greg Mirsky: But if have rerouting, will be same as SRv6 (?) Greg Mirsky: Please join PALS session with Open Design Team - it may not be what you think it is. Lou Berger: Out of time, please continue the discussion on the list. It's clear we need an interim on this topic. Xuesong Geng: (?) ## #11 14:22 9 min Title: DetNet Enhanced Data Plane ### Presenter: Li Zhang ### Draft: https://datatracker.ietf.org/doc/html/draft-yzz-detnet-enhanced-data-plane-01 start at 1426 Lou Berger: At the last meeting Dhruv requested to have separate requirements and (perhaps multiple) solutions documents. Have you given any thought to this request? Li Zhang: We can consider after. Quan Xiong: (?) ## #12 14:31 9 min Title: Deadline Based Deterministic Forwarding ### Presenter: Shaofu Peng ### Draft: https://datatracker.ietf.org/doc/html/draft-peng-detnet-deadline-based-forwarding-03 (Start at 14:36) Xuesong Geng: There has been a lot of previous work on this topic, it would be good to understand what is novel in this document. Lou Berger: We look forward to hearing more about this at an interim. ## #13 14:40 9 min Title: MPLS Sub-Stack Encapsulation for Deterministic Latency Action ### Presenter: Xueyan Song ### Draft: https://datatracker.ietf.org/doc/html/draft-sx-detnet-mpls-queue-02 (start at 14:44) Lou Berger: Earlier as I mentioned there was a request to separate requirements from implementation - there may be an opportunity to contribute to such a requirements document (see 11 above). ## #14 14:49 9 min Title: YANG Data Model for DetNet Mapping with Network Slice ### Presenter: Xueyan Song ### Draft: https://datatracker.ietf.org/doc/html/draft-sw-detnet-network-slice-mapping-yang-00 (start at 14:52) Lou Berger: Must coordinate with TEAS working group to make sure this work is aligned with TEAS. Lou Berger: There is a lot of interest here in queueing, we will have more discussion on that. ## 15:00 Adjourn