RTWG Agenda Bashing John Introduced Goals and milestones revision, John Scudder Stewart Bryant: Should we include the loop free as a distinct goal John: Sounds fine, if no other comments, we'll publish to the list for comments GTSM to Standards Track, Dave Myer Dave: Would you use multihop? Ted: We could Dave: You could engineer from many more places Ted: You still have the requirements to do loopback peering John: You can do loopback to loopback Dave: That's not the same thing as multihop, multihop is engineered less than 255 Ted: And I can see that, but it would be nice to have with multiple paths between attached peers. Probably half of our peers have multiple paths, some of which might be multiple hops. Chandra: We ahve a shipping version with multipath, and customers have asked us to keep it in there Dave: Okay, that's good feedback. We couldn't analyze this in any way to convince ourselves that the security properties of this are deterministic. Alex: What is the output of this discussion? Are you going to add multihop? Dave: We need to go back and look at it, and discuss it more. John: Multihop is still in there, just as an appendix, so you're still compliant. Alex: The document needs to be clearer as to the security properties of multihop. Chandra: I'm not certain of the point of multihop? You can set the TTL to anything, so it doesn't matter. Dave: If you use multihop, and you accept packets within that trust radius, you don't know much about where that packet came from. It's a lot weaker thing than TTL==255 on a connected segment. Ted: I dont' understand how something so simple has taken so long to get through. Moving it or removing it, and agreeing to do it later. Dave: Sense of the room, removing multihop to the appendix, as non-normative, would this be a good way forward? Sense of the room: Yes Summary of Design Team Meeting for IPFRR/microloop, Mike Shand No comments from the mic oFIB: Show of hands, lots of support Notvia: Show of hands, lots of support oFIB, Francios Alex: How do you identify the changes you signal in the completion messages? Stewart: Was Alex talking about this new stuff, or the main draft? Francios: The length of the message and the description of the state change. Alex: I would like to a description in a more algorithmic way. Francios; Okay. Loop Free Framework Stewart: Can we take a sense of the room on progressing the loop free framework doc to informational as a WG doc? Sense of the room: Unanimous amoung those who have actually read it.