-opened at 9:04 -jzwiebel appointed scribe -agenda bashing, Marshall has trouble figuring out how to talk into the microphone -AD eval is proceeding "fine" -last-call issued on RAC -active drafts on agenda -need to talk about rechartering -Pekka presents on routing arch. - 3 reviews (1 cross-area), of 03, reference to address architecture mobility impacts - moving on to 04 added summary tables after sections, reviewing summaries intro on how protos fit together. Forwarding protocols changed bi-dir status to be more positive (because already being deployed successfully in some enterprises) - trying to figure out what to say about PIM dense-mode - toplogy diagrams BGP (add line for BGP w/o SAFI) - comment Jerome: SAFI 1 and SAFI maybe should be separate - Toerless: legacy BGP don't use SAFI, good to spell out more details, detail for SAFI1 SAFI2, legacy BGP, wants 3 lines. w/o: 1 : 2 - learning sources thinks PIM-SM w/ embedded RP is best comment: saying best solution doesnt alway work isnt good only interdomain solution with v4 is --- but doesnt work so well comment: that is too strong a solution, "don't work" is too strong comment: what is purpose of slide purpose it to convey the MSDP isn't a great solution comment: why? it works, it has limitations, is totally functional in a certain context. Dino doesn't like "bad fit" too negative for MSDP. Pekka needs to look at it and sound more positive for political purposes. - Marshall: many limitations have been dealt with How many commercial operators are using MSDP? Marshall: many Dino: dont want to send wrong message that this is only a research project... there are production level systems that are using this. Pekka: MSDP in this slide is interdomain Toerless: (need to get) many applications in IPTV RP Configuration -enterprises seem to have more auto-rp/ BSR -no comments - RP configuration Lenny doesn't like auto-RP/BSR wants it to be more 'harsh' need to add 'static RP' - RP redundancy comment: should mention static config already discussed in text shep wants to remove anycast from configuration slide to just here. - host interactions z wants note on RTS/CTS, dino doesn't recommendation from floor, but told to take to rechartering discussion - flooding Toerless wanted explanation of R-to-R vs LAN and wants PIM snooping removed from LAN column. Cisco's RGMP replaced by PIM snooping question: meaning of R-R and LAN columns? -will revise -Tom P talked about AMT - look at URL to see what changed [http://www3.ietf.org/proceedings/06jul/slides/mboned-4.pdf] - added hints about how to refresh - IANA section: getting /16 not realistic so changed to /18 - preventing spoofing in the AMT data message - Pekka commented on spoofing possibilities wants it to be "MUST", need to make requirement more clear - Tom's open source no longer matches spec. Who wants to do it? looking for someone to take over implementation -Dino talked about cisco implemenation of AMT [http://www3.ietf.org/proceedings/06jul/slides/mboned-2/sld1.htm] - 3 implentations. Tom Pusateri (open, Pusateri); Dave Thaler, Microsoft(05); Dino, Cisco(06) - Tom's relay works on Linux, but not the gateway. - 154.17/16 used as anycast - will be on 3 Cisco OS. - wants 2byte header reduction. - Cisco AMT implementation, by next IETF, interoperability with FreeBSD and Microsoft implementation based on -06 drsft, get box to microsoft - close on IANA assigning anycast prefix - Confusion on what anycast address to use in the mean time. Needs an AS number to work. - issue that may not be able to IANA address until this is RFC - Dave Thaler-- not an issue because checksum being done at different level - Pekka wants UDP header checksum. Dino says it isn't practical. thayer say, it isn't worth dealing with. Will it get past the IESG? - Pekka is concerned about using the 6to4 prefix, but has no option. Marshall says he'll find one. - Tom says we could use the relay AS, Pekka agrees. Shep says its bad juju. Lots of wonderment. Pekka says 6to4 deals with this. Discussion about moving to 32-bit AS. Shep says, Path would give you the source. BGP purists will not like it. Building route filters for the multiple AS will have problems but it can be dealt with. - Jean, any ISPs interested in this? dino says large data centers 'blah-blah-blah' are intrigued. Shep says content owners are very interested. Parallel file updates. IP video deployments. - Joel says AMT is to address gap caused by ISPs. Dino explained its a way of pushing multicast to the edge. Pekka says transit providers will like it because it will cause more traffic to be sent to their customers. -Sato, RAC [ http://www3.ietf.org/proceedings/06jul/slides/mboned-6/sld1.htm] No comments. The proposal was to merge with AAA, and it passed with no objects, OK to merge -Sato AAA draft - listed changes in this draft. -ASM question - Sato: independent of ASM/ SSM currently, won't work with ASM tunnel - Marshall wanted to know is this theory or implementation. Sato says there are no implemenations, but NTT will be doing it when the draft is mature. - Marshall asked will it be propiretary or open. - Sato first step is proprietary mechanism but goal is standard McWalter: MIB - little change from last IETF, 2 changes, some additional statistics - reviewed by Pekka - waiting to make sure no changes are necessary due to PIM - toerless, scopes and zones question. Wanted to know about restrictions on scopes - dino, RPF drop counter question are DF failure drops part of this same counter. - Marshall says 'its been around' so moving to last-call. Wei: igmpv3 lite. 'liu' - Excellent ppt summary [http://www3.ietf.org/proceedings/06jul/slides/mboned-0/sld1.htm] -applies to IGMPV2 and 3 -IGMPv2MLDv3 is complex, some functionality not used so much lite and full version should work together - question on whether or not it is worth changing host implementations - z said excellent summary, but we should just do it as a BCP on the router to remove the 'exclude filter mode' - z said if we're going to simplify it, go even further and do just block and allow. Marshall asked if it would interoperate, z said it needs more study but not a problem in general. - Toerless said Cisco said IGMPv3-lite was already in use so change the name. Can we do a RFC on a subset of an RFC. Marshall says some say yes, others say no. - Open Issues -going further on host side if necessary - reducing nmore meesage type if appropriate - Marshall: Magma is going away, mailing list alive but new protocols arent working on - Brian supported BCP. - Marshall says AD said MBONED could move forward. - Marshall: other issue is issuing a partial protocol, Marshall fully supports BCP, could also be informational. - Brian says 'informational document'. - Lenny wants to restate the problem. Specifically settop boxes, settop boxes often dont yet support IGMPv2.. therefore valuable to implement easier solution, supports effort at BCP - Marshall "what can you drop safely". Lenny wants BCP to simplify. - Toerless wants to focus on SSM if we simplify - Marshall will present to the list as a BCP. Koch: mcast.net issues -what is maintenance policy for mcast.net DNS zone? - most of the ip addresses exist - multicast and DNS intersections. Where is it homed? It seems to be a problem. There is no policy. - Joel suggests that mcast.net isn't useful to him any longer. - toerless says ssm-translation is used, but... it isn't that useful all the time. Marshall wanted a written description but Toerless say they don't exist. It's not standardized. - Simon Leinen: likes to see the names in "netstat" output. - Marshall: IANA occassionally allocates new addresses. - Marshall: you can do your own multicast address in an A record. Peter, but it wouldn't be consistent. - Toerless thinks its helpful - Marshall suggests this is going to be painful to figure out.