WG Overview Chairs 10 msdp yang model published as 8916 pim-yang in rfc editors queue igmp-mld-snooping-yang in rfc editors queue dr-improvement returned to WG by AD. null-register-packing returned to the WG. bfd-p2mp-use-case in WGLC igmp-mld-proxy-yang discussed in this meeting igmp-mld-extension discussed in this meeting pim-assert-packing expired sr-p2mp-policy revised back in october igmpv3/mldv2 to internet standard survey done. draft-ietf-pim-igmp-mld-extension Stig Venaas 15 defines extension tlv and a registry Ready for wglc 6 people feel its ready for wglc. draft-ietf-pim-igmp-mld-proxy-yang Hongji Zhao 10 Welcome more comments ready for yang doctor review draft-ietf-pim-null-register-packing Ramki C 10 requested publication and our AD found several issues. Need more reviewers Alvaro: yes, last call it and have discussion which didn't happen last time. No security considerations section, etc. We need the wg to look at them better. Stig: need volunteers to review the document. Sandy: Will review. Lenny: Will review. draft-chen-pim-srv6-p2mp-path Huaimo Chen 15 added details on ingress/transit nodes Acee: is this the first time this forwarding behavior is being described for p2mp tree or described somewhere else? Huaimo: No spring document, this is new. Stig: what is new is trying to define the whole tree in the packet. Acee: doing it at the ingress and process for nested sub trees. Mike: the pim wg has adopted a doc to do something similar and does create state and it's a good solution. now just progressing a more sr like solution which is more similar to segment routing. Stig: the main thing is how efficiently can you do this. in hardware? Huaimo: SRv6 we have flexibility to do these procedures. Mike: it is a valid concern to have a growing sl which is a sr concern. Huaimo: there are some solutions, can use a sub tree. another solution people are working on compression sid's to help reduce concerns. Ian: reinforce that work exampels are provided. skeptical that this is something that can be implemented across hw implementations even with micro sids. Mike: because of the sl size? Ian: Yes. Huaimo: The number of branches are a fixed field and we can then get the sub trees. We can access the sl in the header. the destination is also in the header which will persist. SID's have argument functions and hardware can access those functions. Ian: It would be nice to see some concrete work examples. Greg: second examples, scale and processing. look at a real network and see what the transport tax is in scale. it becomes unbounded at somepoint. we are asking for examples and see what the packet really looks like. Huaimo: the segment list can be smaller Greg: you haven't fixed the problem but have spread it around. Huaimo: work going on in spring for compression for sids. Greg: defining the entire the tree doesn't compress the packet. Huaimo: beauty of sr is the entire path is in the packet. Greg: also service seperation. give us some examples. Mike: good point. part of what Huaimo is describing is general sr concerns. Stig: how much space needed to store this for mtu size. how much overhead to describe a typical tree. how many people have read this draft? 9 people. do you think we should adopt the draft in pim now? 4 to adopt and 8 people against adopting right now. So respin the document to include the examples as requested. Authors send email to list and start discussion. draft-song-multicast-telemetry-06 Mike perhaps develop a problem statement in mboned and protocol mods in pim in coordination with ippm. Add branch identifer to the instruction header. Making telemetry data efficient for multicast. ippm wg would be the ideal place for this work but there is no multicast expertise or interest. Lenny: what exactly was the experience in ippm? Mike: zero response from the wg. busy wg. multicast oam is not a priority. pinged chairs feel free to keep trying. Lenny: you wouldn't anticipate ippm chairs would feel their toes are being stepped on? Mike: no but we will need to check. Stig: Should be in our charter. but would need to check with chairs and ADs. But should first see if there is any interest in this work. Stig: Do you think we should do work in this area? 2 people we should not. Handful of people are in support. Lenny: Is it making changes protocols changes to pim. Mike: not changing any multicast protocol at all. only adding a branch identifer to existing telemetry protocols for efficient multicast tree reconstruction. PBT, H2Step. IOAM. We are only talking about telemetry data for efficient collection. Lenny: very oam related. Mike: yes. Lenny: this would seem more mboned related than pim. Stig: probably better fit in mboned. Alvaro: agree with Lenny. mboned specifically talks about extending protocols for tools. very premature talking about adopting the work. talk to the ippm guys to make sure they are ok. Understand ippm's point of view. Mike: I will email the ippm chairs and copy mboned/pim chairs. Stig: either way a multicast telemetry needs to be done in mboned.