Multimob (Multicast Mobility BoF) IETF-75 Chairs: Gorry Fairhurst and Behcet Sarikaya Note-taker: Mathias Waehlisch Behcet: Presented the agenda and the IETF Note-Well. Gorry: I'd like to emphasise that this is the second BoF, which focuses on charter discussion (not specific proposals for multicast). Behcet: Presented the "Multimob Problem Statement". Dirk von Hugo: Presented "Requirements from a Global Operator". There were no questions/comments. Gang Chen: Presented "Multicast Support Requirements for PMIP6". Rajeev: Considering the first and third bullet (second slide: Bandwidth control + forwarding): Are these the only requirements for the LMA? Gang: We can use encapsulation technology from home agent tunnels, or GRE. Rajeev: Are you transporting the multicast packets only in unicast tunnels? Suresh: There are solutions that do both (two classes of solutions). That is, operate on PMIP as is, and ones that extend PMIP for optimized multicast traffic flow (see the presentation afterwards). Rajeev: Do you want to carry multicast traffic in unicast tunnels? Gorry: The charter suggests to start from this position for the first milestones. It is clearly not the final optimal goal to use individual point-to-point tunnels for multicast. Rajeev: Are you saying multicast tunnels are in scope? Gorry: Yes. Ralph Droms, via Jabber: Are these "handovers" all within a single access technology? I.e., not related to discussion in netext2 earlier today? Gorry: Currently the Charter sees this as within the same L2 technology. Thomas: Presented "Group Management Solutions Overview" There were no questions/comments. Gorry: How many of you have read the IGMP/MLD solution drafts? Suresh: Presented "Proxy Mobile IPv6 Solutions Overview" Rajeev: Does the scope of multimob consider all the different classes of solutions? Gorry: The baseline spec will take PMIPv6 as it is. In the second step, we will consider changes to optimize PMIP multicast distribution. Rajeev: That sounds like a good plan. Why are you concerned about any changes to PMIPv6? What are the problems? Suresh: In the last BoF there was feedback that changes are not really required. Gorry: By changes we mean specifying how this should work at the MAG and LMA, and extensions for multicast. Thomas: Another aspect is that deploying multicast on PMIP is not so obvious. It makes sense to elaborate this. Rajeev: As said before, it seems a good plan, but we should not limit approaches to unchanged PMIP solutions. Gorry: Presented "Charter discussion". Milestones: Rajeev: Regarding the one year milestone: What do you mean with multicast mobility, do you mean handover? Gorry: Fast Handover - Yes. Suresh: The time frame seems quite tight for this second set of deliverables. Gorry (not as Chair): I can see that it may take longer to get the extensions agreed and defined. Jabber question from Jari: "I think it is obvious that we need to do SOMETHING about multicast in proxy MIP. The word "multicast" does not appear in RFC 5213. However, the main question from my perspective is the suitable ambition level. The implementor crowd wants to do as little as possible, the researchers want to do as optimized as possible. My opinion is that we have to learn to walk before we can run. I like the way the charter divides the PMIP work into basic work and extensions. At least the basic is needed. Hopefully the audience can provide input on whether they believe the extensions are needed as well." Gorry: Would people like to speak on the Charter contents and deliverables: Should we break the work in two parts (changed and unchanged PMIP solutions)? What do people think about PMIP extensions? Hitoshi: What does the charter mean with "optimization"? Behcet: Optimization approaches to increase efficiency of multicast traffic delivery, e.g., the avalanche problem. Hitoshi: PMIP extensions are needed to support aggregation and seamless handover. There is the problem that an unsolicited MLD report will not be sent at the new MAG. Thomas: PMIP uses a point-to-point link towards the MN. Attaching to a new MAG results in a general MLD query initiated by the MAG/AR. The main problem is that the distribution exhibits a lot of redundancy: You can have several MNs requesting the same traffic from different LMAs. Gorry: Is the charter well-specified and does it describe useful work for the IETF? How many people believe that the topic is not yet sufficiently well-specified or of interest to the IETF? <0 people> ... feel we should not proceed with this work at this time in the IETF? <0 people> ... Who has read the PMIP documents?: A reasonable number of people... Ralph: I'm interested in judging energy levels: Who is willing to contribute to the overall work of this group, specifically who will comment on drafts and will offer reviews: Albert Tian: Are there any operators willing to qualify the usefulness of the protocol changes? Dirk (Deutsche Telekom): Adding appropriate multicast support in PMIP is important. Gorry & Behcet thanked all the people who put so much energy into preparing for the BoF - including the many drafts, conversations and contributions to this BoF meeting. The meeting closed.