2 drafts published since last IETF. The few people that responded where happy with the LISP join-attributes last call. Yang models, pim and IGMP/MLDP they are complete, but can?t be published. 1. Dave Allan, update to draft framework for computed multicast applied to mpls based segmented routing a. IGP extensions for ISIS and OSPF b. Questions i. cisco, pain of customer is complexity, does this make it any simpler? Ans removing a multicast protocols pim ii. tim chan: whats doing the computation, SDN type iii. dynamic or static environment, what are the trade offs of using this vs not using this. Ans multicast state into IGP iv. need to run this by some of the guys that have some segmented routing background v. some of the questions is how to comput the state for the entire tree vi. Andrew, what is the benefit for introduction of this, cost is large $100,000 and training of people. Concern of not being deployed for complexity Ans> large number of Sparse tree to be simplified, IGMP recovery is faster. vii. Ice: complexity and operational cost of running, this does it really simplify Multicast solutions. viii. Very quiet from comment point of view. 2. Ice, generic multicast router election on LAN??s MLD LAN election a. problem DR/DF election mechanism b. is PIM the right place to have DR/DF selection c. proposed DR/DF election in IGMP and BIER can benefit d. Group base DF election, (note DF and DR election are the same) highest random weight for a particular group is the DF. e. introduce IGMP hello create a neighbor list, f. DA is elected base on the Hello list, DA creates list of Canadidate DFs called DAL, DAL is used to elect a DF by all candidates. g. chair comment: existing draft that addresses this. EVPN draft. BSR hash function. h. driven by BIER mechanism i. questions: i. DT, DA any plans for backup and what principle ANS: any other router is a DA, highest address for IGP hello is DA. ii. DT> what happens when the highest come back is there a revert mech ANS: YES iii. Question: group address selection why not using source ANS: IGMPv3 we can take the source as the key for weighted random iv. Question: introp between this new IGMP/MLD msg and old IGMP 3 ANS: this is a new v. Question: erricson, concern about DA in case of 8.02 LAN when connected to BIER. ANS: no dependency to BGP vi. Question: cisco IGMP snooping backward compatible? Ans Yes it is and will note in the draft vii. Qeustion: Is this draft BIER or PIM ANS: do it in BIER or here? It should be here with dotted line to BIER, as such maybe 2 drafts? 3. extension mldp proxy luise M.Contreras a. Comment>add ssm or asm support in requirement document, b. Comment> upstream chosen should be chosen on unicast protocol or not? In case of multiple paths c. Comment> mobile scenarios requirements? Does it impact the solution? And use cases 4. PIM DR IMPROVEMENTS Sandy a. Comment> discussing/concerns in DR election problems with backbone, Slow conversion of BGP, DR being active before routing table has been converged. b. Comment> Micheal abrham, IPV6 which IP address are we talking for election. Link local address or IPv6 assigned address conclusion> should be same as PIM today as in link local. 5. MSDP YANG Sandy a. Comment> schema each vendor can config this base on the functionality they have on the router. How do you talk to different vendors to ensure all feature are included. ANS> currently we do the minimum common feature set. Some feature that only few vendor support they won?t put it in the Yang models b. Comment> Customer does not want to have a proprietary yang model, the standard should be a super set and not minimum c. Comment> micheal ebrham we should go for the super set. d. Comment> customer a model is defined and should be flexible to add new objects in future without going through the standard 6. Stig PIM IGP EXT a. pim security not as strong as IGMP b. comment> IGP to distribute source distribution, why not c. comment> Why not use SAP? No clear answer d. Comment> Ice what is the main the benefit?