Minutes IETF117: dmm
minutes-117-dmm-00

Agenda
Administrivia & Intro, WG organization & milestones, Chairs, 5 min.

Working Group Drafts
Mobility aware Transport Network Slicing for 5G, draft-ietf-dmm-tn-aware-mobility, John Kaippallimalil, 15 min.
[Sri] logic I/F was in the draft before?
[John] MPLS, etc exists already. proposal is a new type

[Zhaohui] aligned with TS, but there is a another specific document in TS on it(Luis is the author for it: his comments on chat)
[John] method is outlined in the other draft, described here

[Richard Li] two source ports, which one?
[John] tunnel (outside) UDP source port for GTP-U
[Richard Li] different QoS requirements might use same GTP tunnel
[John] PDU connections are mapped to appropriate underlays

[Tianji] Is it the same slice?
[John] same 3GPP slice is mapped to potentially different IP network slice in different segment (N3, N9, etc)

• RTGWG has work on GTP~IP mapping

Individual Drafts
Mobile User Plane Evolution, Tianji Jiang, 15min.
draft-zzhang-dmm-mup-evolution

Tianji explains ANUP. Work to be done in 3GPP but socialized here.

[Sri] what does the author want from DMM?
[Tianji] socialize the work

[Zhaohui] clarification on adoption request. This is informational but believe there is consensus to adopt.
[Sri] what is not clear is what DMM WG should do. Is it an informational document?
[Tianji] yes.

[Satoru] what does this mean to socialize
[Tianji] IETF provides input but is informational. The ideas on wireline part come from IETF (e.g, satellite)

[Zhaohui] this becomes input for 3GPP

[Erik] Based on previous experience, informational is less concerning, question to Peter?
[Peter] It would become a normal document that comes from IETF that is submitted to 3GPP
[Satoru] Which WG will discuss this in the 3GPP as agenda?
[Peter] SA2 is the correct WG for architecture work

[Sri] how many have read the draft?
(6 hands)
[Tianji] initiate discussion on mailing list?
[Sri] yes

ANUP Implementation in 5G with BGP Signaling, Jeffrey Zhang, 10min.
draft-zzhang-dmm-anup5g-signaling

Distributed signaling from an SMF to large number of ANUP
Request to review and provide comments

[Sri] was this presented in the routing groups?
[Zhaohui] no

[Satoru] MUP gateway architecture question, could be a PE, or others
[Zhaohui] this is SRv6 independent

[Hannu] naming of the draft is misleading, it is about mapping of entries rather than anything to do with BGP
[Zhaohui] will make it clear

[Richard Li] BGP is used by transport network operator or mobile operator?
[Zhaohui] there is an existing SRv6 MUP architecture and this is using the same mechanism

Architecture Discussion on SRv6 Mobile User plane, Miya Kohno, 10min.
draft-kohno-dmm-srv6mob-arch

This is for fixed wireless or IoT uses, not a general solution.

Request to consider adoption as a WG document.

[Sri] this is an informational draft?
[Miya] yes

[Sri] how many have read the document?
(6 raised hands)
Need more reviews before adoption.
Suggest to raise more discussion on the list.

Mobility Capability Negotiation as a 5G Mobility Pattern, Tianji Jiang, 15min.
draft-yan-dmm-man

host initiated and network based protocols in MCN

[Sri] where does host capable of MP-QUIC fit in?
[Tianji] wireless device is going to have both host and network capabilties

[Sri] as a capability indication there may be value, but for SIP and others we have lost the battle.
[Tianji] this draft is analyzing and classifying only.

[Sri] questions, feedback?
(0 feedback)
Lets look and apply to current contexts.

Impact analysis from IPv6 GTP-U checksum calculation, Tetsuya Murakami, 15min.
draft-murakami-dmm-udp-checksum-impact-gtpu

many UDP nodes still require checksum (non zero) that affects performance (and packet loss in some cases)
ideally, UDP checksum zero should be recommended.

[Leo] have seen these issues in the field.

[Erik] RFC 6936 and 6935 recommend that checksum zero should be used.
[Tetsuya] This is to inform 3GPP.

[Tianji] is this when congestion happens in the network? N3 bandwidth is usually sufficiently large
[Tetsuya] focusing of the general issue.

[Jeffrey] this is a good reason to do ANUP.

[Suresh] the RFCs were for this kind of problem. The action is for 3GPP and perhaps a liaison should be sent to 3GPP?
[Satoru] 3GPP spec allows checksum zero
[Suresh] A liaison statement would be a good idea
[Sri] agree. Are there any backward compatibility issues to note. This is good work.

Report from the side meeting on Mobile Traffic Steering, Marco Liebsch, 10min.

various scenarios of interest for traffic steering

[Linda] when steering UE traffic, is it based on processing time, service delay, ...?
[Marco] based on various constraints and policy
[Linda] document in IDR for reactive selection using routing plane (5g edge metadata)

[Sri] for end marker, do we need more work?

[Richard] Network side is fairly complicated after mobile anchor. Why is segment from gNB - UPF not considered?
[Marco] attempt is to not interfere with 3GPP work
[Richard] 3GPP only defines an interface, not in between

[Satoru] SSC-Mode 3 is the default solution for UPF relocation in 3GPP.

RADIUS Attributes for 3GPP 5G AKA Authentication Method, Sri Gundavelli, 10min.
draft-gundavelli-radext-5g-auth

AOB
Encoding 3GPP Slices for Interactive Media Services, Tianji Jiang, 10min.
draft-jiang-tsvwg-slice-media-service

XRM service that requires low latency and high thruput
(new SST - HDLLC)

For encrypted traffic various IETF classification like DSCP is not sufficient. Propose to use UDP option (tsvwg draft)