IETF 88, Netext WG Meeting Minutes (1) Charlie P. taking minutes Behcet S. jabber scribe WG update - Update notifications in AUTH stage - Prefix Delegation: finished WGLC, new revision (...-12.txt) addresses last call comments - Separation of CP and UP for PMIPv6 - There are some new use cases - Marco: QoS draft version 05 is published. Need more WG review - JC Zuniga: need to have meaningful feedback on the LIF (logical interface) draft. One author has not even updated draft to show changed affiliation ... troubles with projector / laptop interface Proxy Mobile IPv6 Extensions to Support Flow Mobility 15 Mins draft-ietf-netext-pmipv6-flowmob-08 Carlos Bernardos - WGLC with ...-08.txt - Not too many people have read the draft... - Behcet: The authors have not responded to my comments - Brian: Authors did send email responding to Behcet's comments - Brian and Behcet are arguing...? - Rajeev: Can't take meeting time looking up email extracts - Behcet: All of my comments are technical but we are speaking on two different frequencies - Rajeev: Please talk to the authors face-to-face - Rajeev: Looking at call-flow diagram Why does it use the already existing UPN (update notification) messages? UPN was designed for something completely different. Does the implementor have to implement UPN RFC? - Carlos: We just did what the working group consensus asked to do. Civic Location ANI Suboption for PMIPv 10 Mins draft-pazhyannur-netext-civic-location-ani-subopt-00 Rajesh P. - RFC 6757 defines ANI option for PMIPv6, with various options - Getting geolocation indoors is challenging, but Civic Location is a valid option - Proposal: new sub-option for Civic Location - PMIPv6 on WLAN controllers, or on WiFi Access Points - ANI behavior depends on deployment = provided at initial session establishment = client mobility between APs may not be reported to LMA, and thus might be stale = proposal: ANI update-frequency can provide hint to LMA about how often it updates. MAG on AP may use TTL of FFFFFFFF MAG on AP may use TTL depending on how often it sends PBUs - Charlie: the frequency cannot be anywhere near constant, so the terminology may be wrong. Plus, the need for a new ANI depends on the direction of movement - John K.: Read the draft in detail, and it provides a needed functionality - Rajesh: When client moves from one AP to another, MAG may have to re-initiate session. - Deng Hui: Usually use VLAN to manage mobility. But with multiple APs on same VLAN, ANI could be stale? - Kent Leung: MAG telling the LMA? Better: don't need to send updates unless something happens? - JC Zuniga: How do you provision the Civic Location? Are there privacy considerations? Answer: the privacy considerations are about the same. - JC: Don't understand the use case... Answer: people are already doing this today - Rajeev: How many people think this is an interesting idea? Annswer: some hands raised, not a lot. Mapping Wi-Fi QoS in a PMIPv6 Mobility Domain 15 Mins draft-kaippallimalil-netext-pmip-qos-wifi-03 John K. - Numerous changes - New co-authors - ...focussing specifically on WiFi case - Why we need per-user QoS and what is missing... - Gap: how to signal QoS in WiFi access when MN initiates request - Gap: How to map WiFi QoS parameters to PMIP QoS - Rajeev: (on Case A:) when the default bearer is set up, already have admission control and QoS, right? Ans: no signaling between MN and MAG Q: so you are assuming this is *after* default bearer is set up - Sri: This is when an application dynamically issues a request - Deng Hui: When doing AP authentication...? - Rajeev: Scope is that you are trying to map PCC QoS to WLAN QoS - Rajesh: Scope is more general. QoS draft earlier... if there is some signaling, and WiFi between MAG and MN, the QoS between MAG and LMA may not agree with QoS between MAG and MN. - Rajeev: in that case, should get rid of language like "default bearer" - Sri: That language is not in the draft... - Sri: UPN message can come back after ADDTS (??) - Case C: Hybrid: can skip PBU PBA - Rajeev: Why do UPN and UPA if you are not going to complete the sequence - Kent Leung: both ends have to have QoS - John K.: will explore the most optimal way - Mapping of Connection Parameters: = TCLAS --> PMIP mobility session = 802.11 QoS TSPEC --> 802.1D = Mean data rate <= GBR = Pre-emption priority (ARP may be used) (admission control) - Marco: don't understand the justification E2E QoS without Admission Control - Use DSCP - 802.1D UP defined by GSMA and RFC 4594 - Alternatively, use QoS MAP for different mappings per user. Sent by LMA - Rajeev: Working group seems to think this is a good idea - Rajeev: The draft used to be mapping EPC to PMIP, but now it is broader. This needs to be made clearer in the draft - Rajesh: help client mark DSCP packet - Rajeev: For clarification, this is best practice. There isn't any new protocol proposed. - Rajeev: Should the draft be taken up as a working group item? no objection: will confirm on the mailing list EAP Attributes for WiFi - EPC Integration 10 Mins draft-ietf-netext-wifi-epc-eap-attributes-04 Rajeev K. - Following recent progress in SaMOG - In 3GPP, single PDNs versus multiple PDNs - PDN type is either v4 or v6 - S2a PDN connectivity vs. NSWO - Shared new draft with IETFers that attend 3GPP - Clarify single connection mode versus multiple connection mode WLCP - Add PCO attribute - Clarify text around DHCP - it is not used as a trigger Separation of Control and User Plane for Proxy Mobile IPv6, 15 mins, draft-ietf-netext-pmip-cp-up-separation-00 Charlie P/Ryuji W. - Applicability of Proxy Mobile IPv6 for Service Provider Wi-Fi Deployments, 15 mins, draft-gundavelli-netext-pmipv6-wlan-applicability-06, Byju P. - Motivation: momentum around use of PMIPv6-based architecture for SP WiFI deployments - Flat model (EPC integration) versus Hierarchical (domain chaining) - WLAN Access Network Types: controller-based versus autonomous - Deployment Considerations - Flat Model Deployments: CAPWAP - PMIPv6 - RADIUS (to both billing and PCRF) - Hierarchical Model Deployments IETF 88, Netext WG Meeting Minutes (2) Minutes by Juan Carlos Zuniga. Rev: 0 Network-Based Mobility Extensions (NetExt) WG meeting MONDAY, November 4, 2013 1740-1940 Afternoon Session III ----------------------------------------------------------------- 1. Logistics (Bluesheets, minutes takers, jabber, agenda bashing) 5 mins 2. WG Status update Chairs 5 Mins Carlos Bernardos: last version addresses IETF last call Marco: qos draft. We got a couple of reviews. A revised draft has been produced. LIF: Juan Carlos: the document has been stalled for a year or so. What is going to be the next step? Chairs: Raj was the handling this. Will ask him about how to proceed next. Marco> QoS ... Juan Carlos > ... 3. Proxy Mobile IPv6 Extensions to Support Flow Mobility 15 Mins draft-ietf-netext-pmipv6-flowmob-08 Carlos B. Behcet: None of my comments have been addressed Brian: As shepherd; I’ve seen response to the comments Carlos: I have replied Behcet: I will bring strict comments during WGLC Rajeev: we cannot show list comments Behcet: I have not seen my comments addressed one by one Rajeev: Please address comments during this meeting, as we need to move on Rajeev: I attended remotely Berlin and I didn’t get the feeling that using UPA was an agreement Carlos: We discussed and since the new functionality was now available, it made sense to use them Juan Carlos: I remember we discussed issue by issue and no one opposed the resolutions proposed by Carlos Rajeev: Will look at the mailing list and compare 4. Civic Location ANI Suboption for PMIPv 10 Mins draft-pazhyannur-netext-civic-location-ani-subopt-00 Rajesh P. Charlie: I’m surprised, because frequency of movement between APs is not constant. Rajesh: we should probably change the name to reflect what it means CP: Direction of movement matters Charlie: .. John K: I have read the draft and I like the idea. Deng Hui: Where is the controller? R: It may not be on the data plane. If the applications do not the information immediately, the MAG can update periodically Ken:... Juan Carlos: Could we be opening a privacy issue here? Ken: Probably similar to geopriv 5. Mapping Wi-Fi QoS in a PMIPv6 Mobility Domain 15 Mins draft-kaippallimalil-netext-pmip-qos-wifi-03 John K. Rajeev: In the control bearer you get QoS? John K: You get the default bearer done with best effort Rajeev: you are trying to map EPC QoS to WLAN QoS. EAP success resolved on profile of UE which goes on QoS. Rajesh: The scope defines mapping beyond EPC. Case A is not really a 3GPP case, but Case B is. Behcet: In netext we should not provide mechanisms to change MN behavior Marco: This is not a specification, right? John K: So far we are not specifying, but only providing recommendations, like GSMA mappings. Rajeev: are there objections to move the work along and adopt as best practices? Will confirm on the mailing list 6. EAP Attributes for WiFi - EPC Integration 10 Mins draft-ietf-netext-wifi-epc-eap-attributes-04 Rajeev K. 7. Separation of Control and User Plane for Proxy Mobile IPv6, 15 mins, draft-ietf-netext-pmip-cp-up-separation-00 Charlie P/Ryuji W. Rajeev: What is the router doing? You should state that there is a trust relationship between LMA and Router Sri: ... Kostas: You need to specify the functionality between LMA and router. Juan Carlos: The blue line to the right of the LMA to the internet is conveying what? Are you anchored at LMA or router? Rajeev: it could be PCRF policies Juan Carlos: Then you should change the colour… Marco: The scope needs to be clear, and the signaling specified. Not the architecture. Sri: When Alt-CoA was discussed we should have had similar discussions. Praveen: The separation is giving the ability to do load balancing. Rajeev: You can also have one engine dedicated to CP or DP, get PMIP messages on one point, and GRE tunnels on another point. 8. Applicability of Proxy Mobile IPv6 for Service Provider Wi-Fi Deployments, 15 mins, draft-gundavelli-netext-pmipv6-wlan-applicability-06, Byju P. 9. Summary and Next Steps 5 Mins, Chairs