Multimob (Multicast Mobility)
IETF-81
Chairs: Stig Venaas and Behcet Sarikaya
Note-taker: Dirk von Hugo


The MultiMob WG had adopted the draft on IGMP/MLD parameter tuning as WG I-D and will use the current version as starting point for an improved performance using IGMPv3 and MLDv2 protocol for mobility, aiming towards a guideline for query and other timers and values tuning. The draft describes three deployment scenarios for IGMP/MLD message exchange:
§      directly between host and access router on a wireless network,
§      between host and home router through a tunnel link, and
§      between home router and foreign router through a tunnel link.
According to advice by Jari as AD the document should be rather informational instead of BCP which needs a lot of implementations to achieve the status.

For to solve issues in Tunnel Convergence (between MAG and different LMAs for different MNs) several solutions on different problem views (such as business vs. IP network/provider level, e.g. local content distribution) are under discussion:
§      Direct routing: "PMIPv6 Multicasting Support using Native Infrastructure"(draft-sijeon-multimob-direct-routing-pmip6-01) proposes a local routing approach making direct connection between MAG and multicast router.
§        Multicast LMA: "Support Multicast Services Using Proxy Mobile IPv6" (draft-zuniga-multimob-smspmip-06) recommends to use a multicast tree mobility anchor as topological anchor point for multicast traffic, while MAG remains as an IGMP/MLD proxy. This enhancement to the base solution RFC 6224 benefits from reducing multicast traffic replication and supporting different PMIPv6 deployments scenarios.
Different use cases such as access to content in home network, local content, or route optimization independent of where content and MN are located exist. Consensus is that no approach solves all problems - a question whether to concentrate on most likely deployment scenario. For this no formal requirements draft has to be compiled but a clear view on the different assumptions and constraints should be achieved on the ML (mailing list).

The Fast Handover support solutions presented cover four different drafts
§      "Multicast Context Transfer for Mobility Protocol Standards" (draft-schmidt-multimob-fmipv6-pfmipv6-multicast-04) attempts to optimize for handover for seamlessness by synchronizing multicast with unicast operations, shall be aligned with mobility protocols via transparently plugging in multicast of all flavors, tries to abstract from multicast routing in terms of remaining independent of routing solution, and simplifies protocol extensions using existing message structures.
§      "Context Transfer Protocol Extension for Multicast" (draft-vonhugo-multimob-cxtp-extensions-00) builds on explicit tracking of MN multicast status at MAGs and transfer of that information between old and new serving MAG to achieve lossless forwarding of multicast data to a mobility unaware MN. New messages to the context transfer protocol for multicast (M-CTD) announcement and replies requesting buffering, forwarding and leaving multicast group at old MAG are specified.
§      "Fast Handover for Multicast in Proxy Mobile IPv6" (draft-hui-multimob-fast-handover-03) specifies a predictive fast handover mechanism to reduce handover latency and packet loss in Proxy Mobile IPv6 Multicast by specifying extensions for Handover Initiate (HI) and Handover Acknowledgement (HAck) messages in terms of M flag indicating Multicast option and the MN multicast status (group addresses).
§      "Rapid acquisition of the MN multicast subscription after handover" (draft-contreras-multimob-rams-01) extends the current PMIPv6 signaling protocol by including new multicast information option to update PMIPv6 entities during registration and de-registration processes, as well as new messages to trigger the transfer of such multicast information. Instead of context data between MAGs the information is exchanged via the PMIPv6-specific MAG-LMA tunnels.
During the corresponding charter item discussion a differentiation in PFMIP-based, CXTP-based and speed-up MultiMob base solution approaches is made -  different objectives are to allow for rapid context transfer during homogeneous Handover , context for other purpose (e.g. heterogeneous handover), and the synchronization of uni- and multicast traffic.

Multicast source mobility is addressed by two proposals:
§      "Mobile Sources for the PMIP base solution" (draft-schmidt-multimob-pmipv6-base-source-00) explains operations of base multicast support in PMIPv6 for mobile multicast sources with minimum extensions at the cost of potential suboptimal routes (simplicity vs. optimization)
§      Multicast Source Mobility Support in PMIPv6 Network (draft-zhang-multimob-msm-03) proposes to introduce Multi-Upstream Interfaces MLD Proxy (MUIMP) to solve Disadvantages of multiple parallel MLD Proxy functions running on one MAG which is required in absence of multicast forwarding information base (MFIB) at MAG. In case LMA is not operating as Multicast Router further extensions are required.
Due to lack of time the latest - not yet chartered - topic could not be discussed in more detail. All issues will be extended on the mailing list to achieve convergence until next meeting.