IETF 92 - MANET WG Notes o WG Status/Overview [15] - Recap of Documents/Status - Announcements - MIB Documents Bob Cole - biggest problem is there are no comments, last call, they pass, but get blasted by MIB Doctors (w.g. 100 comments). New thing DisMan with not a lot of experience, etc. Need help on this from DisMAN MIB doctors (e.g., Randy Preshun). Adrian Farrell - MIB doctor review early is a good idea. After this week I might be a resource to review it. There is a mailing list for contacting the MIB doctors. Bob Cole - thinks the current draft is pretty stable. The MIB is not going to change, so it's ready for MIB Doctor review. - Directional airtime metric draft Emmanuel Baccelli - Second implementation of it from RIOT (along with NHDP implementation). Soon time to freeze the spec and to get experimental results. Do you have a path in mind towards this? Justin - I would like to have it started before the next IETF. Before going to LC, we have three implementations, I would like to do a mini-interop. o draft-ietf-manet-dlep: Ratliff [20] How many have read the document? A small minority have read. But there was consensus that it's ready for WGLC. Henning Rogge via Jabber - has been using DLEP-07 and thinks it's ready Rick Taylor - knows of several -07 implementations, ready for -08 so thinks it's ready John Dowdell - thanks the authors for their effort. Very pleased with the document. o draft-ietf-manet-aodvv2: Perkins [20] Stan Ratliff - Can we summarize what's left more quickly? There will likely be questions at the end of this and need time for open discussion. Adrian Farrell - Owes a response on the list about Security Discussion. I like the text in the new draft on security. Stan - wrt security, Chris Dearlove has some recommendations. The author team has taken an action item to go back and craft better verbiage on the concerns. Would like to have an early security review. Jiazi Yi via Jabber - What about the implementation of AODVv2? Is there any interop event to prove the interoperability? Charlie Perkins - Have not been focusing on that, but on issue resolutions Justin Dean - history. What do we do with AODV2: Experimental, drop from WG, etc. If the timeline is followed, getting document ready for WGLC by this IETF. Who has read the -07 AODV2 draft - 4 hands Justin - So not a lot of eyes on the document. Emmanuel - know of an implementation, but not sure how complete. Stan - Of the reviewers, who believes it is ready for last call? 2 hands (both on the author team) Justin - do _not_ think it's ready for last call. Security not well addressed, interoperability w/ Internet, bidirectionally requirements not well addressed ; seems experimental. But the document _is_ much improved over previous documents. So, there is momentum on the document, but not sure. Stan - experimental, kill, or last call by this IETF. Don't think we should kill it. The document has been improved significantly. Charlie - message were to be received only from neighbors ; so you have to know it's a neighbor is the ambiguity. This language has been in the document for some time. Setting TTL=255, enables knowing if the message is from a neighbor. Justin - his concern is on the overall state of the document, but not about _one_ specific issue Charlie - point noted, but feels we're ready for last call. Rick Taylor - I have seen the rate of change in the document. Improvements are now coming fast and furious, but not ready yet. It would be a mistake to kill it now, but not ready for last call. Perhaps give it another 3 months. Jiazi (via Jabber) - If I remember correctly, Adrian drew a line in the sand at the last IETF. Now we are at that line. I think there are still several major issues to be solved, including security, gateways, different options, interoperability, etc. If I remember correctly, the WG consensus was that MANET WG would require industrial useable implementations for standard-track documents. Stan - disagrees with that. No qualifiers on implementations. Emmanuel - See two ways to go: Stick to the goals we've set for ourselves or weigh progress and thinks it's going in the right direction and the rate of improvements is good. John Dowdell - enormous work and discussion over the past 4 months. It's a big document. Please give it a fair review. Please give it a chance. Rick Taylor - if the chairs and WG is OK, but do not relieve the pressure. Should be given a _last_ chance. Stan - the issue is not 100% up to the working group, but the ADs have some say. Has received a comment that they can see an end to the issues where they couldn't before; "not there yet", but progressing so shouldn't kill. Adrian Farrell - anyone here work on software? Yes Anyone work on software that delivered late? Yes. Adrian very heartened about the progress ; Believes it's now readable and reviewable ; Now people are going to do that and now seeing constructive comments. Not ready for WGLC, but can see "an end to the tunnel" Charlie - I really think we can go to WGLC _before_ next IETF. Adrian Farrell - so "prove it", but also by presenting a solid, specific plan. Charlie - have done that with tracking of issues. Have shown responsiveness to comments in short order with volume of recent comments. Adrian - need to identify the intermediate milestones before WGLC Rick Taylor - implicit contract on putting these milestones. This has worked for DLEP. But the WG has a responsibility to assist this by reviewing and commenting. Stan - WG silence is taken as "acceptance", so WG needs to be responsible. Adrian Farrell - if you make your schedule public, people can see what's going and know when to review / comment, i.e., like the purpose WGLC serves. Justin - kill it, experimental, or go one more IETF. It looks like the consensus is to go one more. Make it Experimental if they don't make the deadline. Alvaro Retana (new AD) - make a work plan and stick to it. Would like to see the WG set more specific milestones. Get together as WG _before_ next IETF. Charlie - for example, we could schedule an interim teleconference. The authors have been doing this already. Alvaro - suggestion is that the _WG_ does this, not just the author team. Stan - co-chairs will take action to schedule interim meetings before IETF 93. Jiazi Yi (via Jabber) - did we get an answer on meeting consensus? Stan - Is the plan (go one more w/ interim meetings) the way to go? Consensus was affirmed. Jiazi (via Jabber) would like this to be confirmed on the email list. o Recharter Discussion Justin - what would people like to do? Rick Taylor - many ideas with respect to DLEP Charlie - reduced forwarding backbone, work arounds for where routes exist, hybrid protocols have some promise. Jiazi - consensus must be called on mailing list Brian Adamson - MANET multicast, DLEP extensions. Rick Taylor - +1 on multicast Charlie - multicast and multiple gateways Bob Cole - +1 on multicast and multiple gateways Brian Adamson - also multicast group management and inter-domain multicast, CDS relay selection for multiple interfaces. Ron in't Velt - more work on multicast and link metrics for DLEP needed. Rick Taylor - heterogenous networks, draft about hybrid routing protocols, doesn't seem to be correct approach. One could come up with something clever in this area. Stan - need to charter work is identifiable and finite. Adrian - would love for the WG to look at deployments of its existing protocols to see what the users of the protocols are missing. Cross-fertilize with GAIA IRTF group; one of their "hot uses" is OLSR. Emmanuel - Agree with Adrian. Could be interesting to interesting to revisit what kind of link layers work in MANETs now. Henning (via Jabber) - multicast for mesh. Justin - what about auto configuration? Bob Cole - there is other work in other working groups on auto configuration and management that is becoming MANET friendly. How do we track that from here? Charlie - don't want to diminish the importance of configured and managed networks but MANET should be able to operate with minimum configuration. Wr2, auto configuration, there is work from 10 years ago that could apply even though the "autoconf" group failed to function. Stan - You have used the dreaded "a" (autoconf) word ; It would be insane to do the same thing over again expecting a different result. No and "heck no" to bringing autoconf into this working group. Justin - but we have DLEP now. Adrian - There are 2 related autoconf efforts at the moment. It's more constructive to throw your food at them than have them come here. These include "Homenet" and a new group called "Anima" that is more broad. Bob Cole - I'm not suggesting _auto_ configuration. I think there is work we need to do in plain old configuration. There is work on configuration/management protocols that won't work over MANETs. Charlie - There is a lot of interest in other kinds of routing protocols that resemble MANET, but instead are "tree based", where one node acts as the root for routing and other nodes have upstream/downstream routing. Former "pulsar" routing proposal. This type of routing deserves attention. Resembles what is done in ROLL and 6LOWPAN. Stan - this is a great discussion, but all of you at the mic, please take action and bring these thoughts to the mailing list. Rick Taylor - How do you inform your applications that your MANET is doing something strange under the hood. Stan - E.g., DLEP brings information up to layer 3; what about bringing this up to Layer 7? Emmanual - wr2 auto configuration: On Homenet, they are about to waste a lot of time, taking a well-known turn in the conversation about transitivity, etc. There was an autoconf document that talked about this ; i.e. the baseline of understanding of the semantics of a multi-hop wireless network. Thinks the draft has value, but not sure where its "home" should be. Justin - ideas likely too much for one WG. We need to identify what is in the scope of MANET, but also should recognize that good ideas may need a home in other WGs, etc. Henning Rogge via Jabber - Can the list of idea Justin compiled be put on the list?