Session Date/Time: 21 Mar 2022 09:00

dmm

Summary

The dmm (Distributed Mobility Management) working group met at IETF 113 for a hybrid session. The meeting covered updates on existing working group documents, including SRv6 Mobile User Plane and Mobility Aware Transport Network Slicing for 5G, as well as several individual drafts. Key discussions revolved around resolving outstanding comments for documents approaching Working Group Last Call (WGLC), exploring new architectural proposals for 5G/6G user plane evolution, and addressing challenges in satellite communication networks from a Layer 3 perspective. Audio issues for remote presenters were a recurring challenge during the session, eventually resolved by switching browser clients.

Key Discussion Points

SRv6 Mobile User Plane (draft-ietf-dmm-srv6-mobile-user-plane)
- The document underwent Working Group Last Call (WGLC) in April last year.
- Comment Resolution: Joel Harper's comment on sections 5.1-5.3 (SRv6 endpoint behaviors) was addressed by clarifying these sections as informative. Editorial changes requested by Jeffrey were also incorporated.
- 3GPP Reference Architecture: The text was updated to ensure clarification that the document uses the 3GPP 5G reference architecture for context but does not alter 3GPP specifications, and the mechanisms defined are generic.
- Authors recommended a second WGLC due to the significant changes.
- Decision: The chair confirmed plans to issue a second WGLC post-IETF 113 to allow reviewers, especially those who provided substantial feedback, to re-evaluate the changes.
Mobility Aware Transport Network Slicing for 5G (draft-ietf-dmm-5g-network-slicing-mobility)
- Updates from version 02 to 03 primarily focused on improving the clarity and focus of the draft.
- The introduction was revised to define 3GPP slices as a "logical chunk of 3GPP network resources" and highlight how the draft addresses IP transport network capabilities for these slices.
- The text was made "crisp" and focused on mapping 3GPP network interfaces and slicing within the transport network.
- Remaining actions include sharing with other working groups and seeking more detailed reviews within DMM.
Mobile User Plane Message Encoding (draft-murakami-dmm-mup-message-encoding)
- Several revisions (01 to 05) introduced an ADM message, a sequence number for GTP-U messages, and security considerations.
- The document reports on RFC 8754.
- Authors indicated the document is stable and requested a Working Group Adoption Call.
- Decision: The chairs will initiate a working group adoption call on the mailing list.
SRv6 Mobile User Plane Architecture for Emancipation (draft-sekiguchi-dmm-srv6-mup-architecture-for-emancipation)
- Main updates from version 00 to 03 included simplifying the architecture description, particularly dropping AF policy involvement, and focusing on segments instead of node roles.
- Clarifications were added for inter-work and direct segments, including how MEP segments can interweave and accommodate both types.
- Tianji inquired about the MUP controller's integration with 5GS (as an NF or AF). Satoru clarified that the MUP architecture is independent and provides an API, with integration into specific mobility architectures being out of scope for this document.
- Next steps include further clarification in the next revision and triggering more discussion on the mailing list to prepare for an adoption call.
Mobile User Plane Evolution (draft-geng-dmm-mup-evolution)
- This presentation covered two informational drafts: one providing an overview of 5G user plane distribution trends, and another proposing potential 6G integration of gNB and UPF functions.
- The 6G proposal aims for a simplified, flattened network architecture, optimizing signaling and data planes, and leveraging virtualization. This is intended to be socialized within IETF before potentially being brought to 3GPP.
- The chair reiterated DMM's charter is to use IETF tools, not alter 3GPP architecture.
- Discussion Points:
  - Hanu noted that co-location of I-UPF with gNB is already possible in 5G but moving the PSA lower could impact multi-access scenarios (non-3GPP) and security (exposing subscriptions to access nodes).
  - John raised concerns about QoS, access control, network sharing, and potential impacts on mobility management.
  - Dean supported the proposal, seeing it as a simplification of routing that aligns with CUPS architecture, keeping the control plane centralized while distributing data plane exit points to pure IP.
  - Luis suggested looking at radio functional splits for similar centralization benefits.
  - Hanu re-emphasized technical reasons for centralized multicast UPF, such as PDCP sequence number synchronization across multiple gNBs, which complicates handover if decentralized.
5G Distributed UPFs for 5G Multicast and Broadcast Services (draft-tianji-dmm-5g-mbs-dupf)
- This draft proposes applying Distributed UPF (DUPF) concepts to 5G Multicast and Broadcast Services (MBS).
- Challenges identified include additional provisioning for new C-plane/U-plane functions, complexity from three new interfaces (N6mb, N19mb, N3mb) with mixed unicast/multicast underlay, and inefficiency of centralized MB-UPF in distributed 5GS deployments.
- The proposal suggests deploying UPF/MB-UPF closer to the gNB to simplify N19mb and N3mb tunnels. Long-term, integrate regular UPF and MB-UPF.
- Hanu questioned the decentralization of MB-UPF, citing the technical need for centralization to synchronize PDCP sequence numbers for multiple gNBs. Tianji suggested that distributed MB-UPFs could still manage multiple gNBs.
Satellite Communication: Network Problems and Solutions from L3 Perspective (draft-lin-dmm-satellite-network-problem-statement and related drafts)
- Lin presented a suite of drafts addressing L3/L4 solutions for large-scale LEO (Low Earth Orbit) constellations, focusing on internet access and NTN integration with 5G.
- Problem Statement: LEO networks have a huge number of nodes, inter-satellite links (ISL), global coverage, and regenerative modes (satellites acting as network nodes). Challenges include constant ground station-satellite link flipping (5-10 min intervals), frequent link metric changes, unsteady ISL, and polar area link interruptions, leading to massive IGP flooding and reduced service time.
- Semantic Address: Proposed a 3-index system (shear, orbit plane, satellite index) for addressing in LEO networks.
- Interactive Routing: Proposed a new interactive routing approach to minimize routing overhead, leveraging predictable satellite positions to pre-compute topology and link metrics. It uses "instruction lists" instead of segments, with semantic address compression for less overhead.
- The chair asked for clarification on the relevance of this work to the DMM working group's charter, especially regarding mobility. The author indicated that other groups suggested DMM as a potential venue.

Decisions and Action Items

SRv6 Mobile User Plane (draft-ietf-dmm-srv6-mobile-user-plane): The chairs will issue a second Working Group Last Call (WGLC) post-IETF 113.
Mobile User Plane Message Encoding (draft-murakami-dmm-mup-message-encoding): The chairs will initiate a working group adoption call on the mailing list.

Next Steps

SRv6 Mobile User Plane: Authors and reviewers should actively engage in the upcoming second WGLC to confirm resolution of issues.
Mobility Aware Transport Network Slicing for 5G: Authors to continue seeking detailed reviews within DMM and share the draft with other relevant working groups.
SRv6 Mobile User Plane Architecture for Emancipation: Authors to further clarify the draft and trigger discussions on the mailing list to build consensus for a future adoption call.
Mobile User Plane Evolution: Authors and interested parties to continue discussions on the mailing list, particularly on the implications for QoS, security, multi-access, and the relevance to DMM's charter. John (via email) to post more detailed comments.
5G Distributed UPFs for 5G Multicast and Broadcast Services: Authors to refine aspects of shared and individual delivery modes, and explore control/signaling for underlying transport (VPN, SRv6) and UPF exposure capability. Discussions, especially regarding the centralization debate, should continue on the mailing list.
Satellite Communication: Network Problems and Solutions from L3 Perspective: Authors and working group members to discuss the relevance and potential scope of this work within the DMM working group on the mailing list.