Meeting Minutes from 1.) Carlos Jesus Bernardos Cano 2.) Lyle Bertz ------------------- - Agenda bashing - Active DMM WG documents Chairs update on the status of the different active DMM WG documents: * draft-ietf-dmm-ondemand-mobility-13 * draft-ietf-dmm-deployment-models-03: authors does not react fast enough to move the document forward. * draft-ietf-dmm-fpc-cpdp-10: great work, but need more support/reviews from the WG. * draft-ietf-dmm-distributed-moblity-anchoring-08: Carlos added as co-editor, almost a reset. Needs reviews. * draft-ietf-dmm-srv6-mobile-uplane-01: few discussions, there is a slot today. - Active MIPv6/PMIPv6 Maintenance WG Documents * draft-ietf-dmm-4283mnids-07: should be now fine, found a way to clear the DISCUSS. - 2 LS received. First one responded on Feb 2. The second is still unresponded. Suresh (AD) has not received the second one. Satoru, with 3GPP hat on: new WI in CT4, expecting the chair to do send it to the IETF, already created from 3GPP side. - FPC update. Lyle Bertz presents slides. Authors request feedback. Sri asks for implementation update. Almost 2 years with running code. 2 implementations available using ONOS. Lots of the changes in the draft come from the feedback from the implementations (that are based on version 4). Now that the document seems stable, they might updte the implementations to version 10. Suresh: not really about the content, but about the form. Are you going to check that the document is compliant with RFC8342? There is stuff to fix. - Distributed Mobility Anchoring Carlos J. Bernardos presents slides and asks for feedback. Marco: provided comments offline using track changes. Document is on the right track, but need some improvement due to the cut. Sri: do one more edit and ask he will ask for reviewers in the WG. - User Plane Protocol Study in 3GPP Satoru Matsushima presents the slides. Charlie: highlighted the term control plane impact, and obviously GTP-U has a long history and it is well understood. If you don't want to have control plane impact, you will have to run GTP-C and there are not many alternatives that can be controlled by GTP-C. Satoru: investigating alternatives to the control plane. Kalyani Bogineni: you might not be restricted by GTP-C. In Rel-15 PFC is enfornced. Suresh: this is gonna get out of hand. Back in LTE times we have ad ependency list between 3GPP and IETF. We need to start tracking this more closely. XXX from Huawei (no badge): for which release is this? Sri: IETF is not selecting any protocol. We just standardize protocols. (discussion I don't get...) Arashmid Akhavain: until different solutions are analyzed, the impact cannot be assessed. Georgeios: (I didn't get the question on one LS) Is 3GPP waiting for any input regarding the selection criteria? Saturo: of course, yes. Charlie: maybe we should use the FPC. Suresh: if we want to select something officially, we can do it. - LISP for the Mobile Network Dino Farinacci presents the slides. Co-author: your underlay can be anything, IP, MPLS. Arashmid Akhavain: (didn't get the question) Tom Herbert: confused about what you propose in the data plane. Is LISP data plane part of this? Dino: we don't want to do IP-in-IP. All ISPs said when doing LISP in 2007: use UDP. LISP can use GTP-U as data plane. Suresh: I agree with the sentiment, but this is not going to happen (referring to work together, doing one spec together by different SDOs). Jari: interesting work. About the collaboration among different SDOs, I don't think we can do all one spec, different things each of them do. - SRv6 for Mobile User Plane Satoru Matsushima presents the slides. Sri: any possibility of working without requiring IPv6? (not sure I got the question) Satoru: we already have code. Suresh: SRv6 does not allow insertion right now. There is a draft specifying some coses where this insertion is safe. XX from Huawei (some guy than before, no badge): (cannot get the question) Tom Herbert: this is using IP-in-IP as default. Why not using UDP encapsulation? Darren Kurzs??: some comment I didn't get. John K: how this would work when you have multiple UPFs? SRv6 as Data Plane for 3GPP N9 Interface Arashmid Akhavain presents the slides. Dave Allen: comments that something is a very bad idea. Arashmid: clarifies it. Fabio M.: Dinos presented LISP control plane with LISP user plane or GTP-U. LISP could also be used with SRv6. Samita: Every single solution is using GTP-C. Optimized Mobile User Plane Solutions for 5G Kalyani Bogineni presents the slides. Satoru: appreciate that you made this document. Joel Harpern: strongly agains the WG adopting or using this as a response to 3GPP. Mobility-aware Floating Anchor Sri Gundavelli presents the slides. Suresh acts as chair while Sri presents. Charlie: there seems to be a lot of signalling everytime a device moves. John K: optimization takes place between the origin and the destination. 3GPP is doing something different. Seems like SDN solution, there is a limit on what you can do, distributed routing. Sri: only for certain cases we do it. Dave Allan: ellimination of anchors when they only do forwarding is relatively easy, but when yoi look into virtualization and state, it's different. Huawei guy with no badge: I didn't get the question. Proxy Mobile IPv6 extensions for Distributed Mobility Carlos J. Bernardos presents the draft. Charlie: there is previous work, like FMIPv6, context transfer that might be considered. Consensus on adopting as WG document, to be confirmed on the ML. Router Advertisement Extension for On-Demand Mobility Wu-chi Feng presents the slides. Philip Earoley (BT): is there any negotiation? Wu-chi: there are options put in the RA and the UE chooses. John K: are you going to look at what happens if the 3GPP control plane advertises one prefix and the RA another one? Lorenzo Colleti: you will get a hard time getting this adopted by 6man. Danny: we looked at PVDs, there are alternatives, including using DHCP. 3GPP prefers the RA alternative. Sri: we will discuss on the mailing list, with 6man chairs and we will find a way. DMM Deployment Models Seil Jeon presents the slides. Charlie: does it make sense to show 5G architecture and show how it maps? Meeting adjourned. -------------------- 1. Title: Administrivia & Intro, WG organization & milestones Note Takers - Lyle / Carlos I-D models - waiting on dependencies I-D dma - Carlos is new editor; needs more reivews after rewrite maintenance I-D 4283 - issue resolved and moved to publication queue Liaison Statement Suresh has not recieved the 3GPP UP LS Satoru (wearing 3GPP hat) is waiting on 3GPP CT4 action 2. Topic Name: FPC update Presenter: Lyle Bertz - Suresh asked to ensure a NDMA (RFC 8432) review in parallel to changes - Examples will be added - New update on ServiceEndpoints 3.Topic Name: Distributed Mobility Anchoring Presenter: Carlos J. Bernardos - Major rewrite to simplify complexity (now 15 pages) - Marco has made some comments that need to be fixed - It was urged that maybe changes occur on github or other locations where changes can be seen 4. Topic Name: User Plane Protocol Study in 3GPP Presenter: Satoru Matsushima - Charie Perkins - If you don't want control plane impact it is hard to imagine a lot of alternatives that can be controlled by GTP-C. (Satoru) Correct, we may seen impact to GTP-C and we have CUPS which will be impacted. The 3GPP study will look at control plane impact. (Charlie) - Should we look at CUPS? (Satoru) - Yes (Kalyani) - Noted that in 3GPP Release 16 there is an alternative to look at the UPF API. (Satoru) - Noted this is a SA2 item. (Suresh) - Is there an LS coming for this? (Kalyani) - IETF can ask in the LS or driven from SA2. (Suresh) - Suggests tracking lists of TS/TR <=> IETF draft. Especially if SA2, CT4 Suresh took AI to work with Satoru, Kalyani and others to talk to SA / CT members for tracking. Ultimately the idea is to make an effort to better track the various activities. (Chunshan) - Will this be used for both 4G and 5G or just for 5G? IETF is not selecting protocols for 3GPP. (Kalyani) Is this only N9 related. (Arashmid Akahvin) - We'll provide feedback that is taken into consideration (Georgios Karagiannis) - How will IETF contribute the criteria? (Satoru) - Providing back I-D and other informmation (Georgios Karagiannis) - Is 3GPP waiting on IETF? (Satoru) - Yes (Charlie Perkins) - We should note FPC's existence to IETF (Suresh) - We can note this in a (Dave Allan) - Clarified dependencies and decisions between 3GPP and IETF 5. Topic Name: LISP for the Mobile Network Presenter: Dino Farinacci (Arashmid) - Agree to work together with 3GPP (Tom Herbert) - It was confusing, was the LISP-DP part of the proposal? (Dino) - Not IPinIP (not entropy) GTP vs LISP - Can use either one; main feature is that we have LISP-Crypto and data plane echo for testing (Suresh) - Generally agrees with cooperation but it will not happen as a single spec. (Arashmid) - Did not propose a joint document. (Suresh) - Notes that 3GPP Stage 3 specs are quite different than the I-D. (Jari Arkko) - Regarding cooperation, a good way of doing this which refers to each other's work a wrong way of doing this which is writing a spec and assuming adoption 6. Topic Name: SRv6 for Mobile User Plane Presenter: Satoru Matsushima (Sri) - Any possibility to operate without SRH insertion? (Satoru) - It is possible but depends upon implemenation (Suresh) - In 6man, SRH does not allow insertion but there is another draft on how to safely handle this. The idea is that the insertion does not result in an ICMP error the host cannot do anythin about (Chunshan) - We have introduced Reflective QoS but it appears in the I-D this is not considered. Is SRV6 used toward SMF? (Sri) - send questions to ML (Tom Herbert) - What about using UDP so that it is compatible with LISP? (Darrin Dukes) - Use of the flow label could help the draft here and LISP does not have this problem. (Johnm K.) - Asked questions on ML and Satoru () - TLIFA in the core and several services in one solution as opposed to others 7. Topic Name: SRv6 as Data Plane for 3GPP N9 Interface Presenter: Arashmid Akhavain (Dave Allen) - Using SID to say this is subject to legal intercept is a bad idea (Arashmid) - It is an identifier for a number of functions in the node. How hidden (or not) the function is up to the provider. (Farooq) - Noted that Dino presented LISP-C can be used with any dataplane. There is no reason why it cannot be used with SRV6. (Arashmid) - Agreed. (Chunshan) - Comment - Basically GTP-C is not being changed and only GTP-U but we can basically use any data plane 8. Topic Name: Optimized Mobile User Plane Solutions for 5G Presenter: Kalyani Bogineni (Chunshan ) - The use of current tunnel is per slice as opposed to PDU Session (Kalyani) - We can refine requirements in I-D to reflect this (Joel Halpern) - Expressed opinion against adoption as it did not address the LS request No time for response 9.Topic Name: Mobility-aware Floating Anchor Presenter: Sri Gundavelli / Marco Liebsch (Charlie Perkins) - The packets from the internet have to go somewhere. The Topology and Location DB is usually in the anchor but this solution appears to require more signaling (John K.) - The optimzation of route is taking place between source and destination. There is some optimization on 3GPP that the I-D should consider. There is a need for alignment of topology databases. (Dave Allan) - The elimination of anchor points is straight forward. When you consider stateful handoff this becomes problematic. We must understand that the business of mobile noetworking in these solutions. (Chunshan ) - Consider SSC Mode 2 & 3 change as well as Mobile Edge Computing.in 5G. 10. Topic Name: Proxy Mobile IPv6 extensions for Distributed Mobility Presenter: Carlos J. Bernardos (Sri) - How many have read document? ~6 (Charlie Perkins) - Some of the preivous work (e.g. Fast Mobile IP) considered feautres such as context transfer. Is this I-D an intermediate point along a general design? Do you agree there may be some modifications in the future? (Carlos) - Yes. (Sri) - Rough poll - How many support adoption of this document as a WG document? ~5 How many oppose? None (Suresh) Send to ML (Sri) Will be sent to ML and if there are no objections on the ML it will be adopted. 12. Topic Name: Router Advertisement Extension for On-Demand Mobility Presenter: Wu-chi Feng (Philip) - UE wants one type of SSC and netowrk is offering another. How does that work? (Wu-chi) - Network can offer colored prefixes and UE chooses (John K.) - 3GPP has a CP that will note the advertisements on the CP on a PDU basis. Are you going to look at address conflict between CP and UP? Further discussion will be taken offline. (Lorenz) - It will be hard to get bits from 6man and you may want to look at PVD WG. It may not do want you want but should be looked at. (Wu-chi) - Authors have discussed this with PVD. (Danny Moses) - Noted PVD was looked at and they looked at DHCP. As an initial response from SA2 the alternatives (PVD and DCHP and others) were noted but SA2 responded with RA being the most promising. We will take this up on ML and talk to 6man chairs. It will need to be socialized with 6man. 13. Topic Name: DMM Deployment Models Presenter: Seil Jeon (Charlie Perkins) - Does it make sense to show some of the 5G architecture and how DMM is embedded in that? (Seil) - This model already meets this.