[14:13:16] --- intvelt has joined
[14:38:28] --- intvelt has left
[19:13:39] --- teco has joined
[19:17:42] --- teco has left
[19:18:49] --- teco has joined
[20:28:36] --- tom5760@gmail.com has joined
[20:33:35] --- yangwoo has joined
[20:41:11] --- yangwoo has left: Computer went to sleep
[20:43:38] --- dthaler has joined
[20:44:53] --- Will Ivancic has joined
[20:45:01] <dthaler> agenda and document status <http://www3.ietf.org/proceedings/07dec/slides/intarea-0.ppt> agenda and document status <http://www3.ietf.org/proceedings/07dec/slides/intarea-0.ppt>
[20:45:10] --- TJ has joined
[20:45:12] * dthaler has changed the subject to: Autoconf WG Meeting
[20:45:35] --- kakima has joined
[20:45:51] --- intvelt has joined
[20:45:52] --- heer.cs has joined
[20:45:59] <dthaler> wrong link
[20:46:01] <dthaler> Agenda <http://www3.ietf.org/proceedings/07dec/slides/autoconf-0.ppt>
[20:46:06] --- TClausen has joined
[20:46:57] <dthaler> o Agenda bashing & Working Group status update - 05 mins
[20:47:12] --- TClausen has left
[20:47:15] <dthaler> next up
[20:47:16] <dthaler> Mobile Ad hoc Network Architecture ID Progress draft-ietf-autoconf-manetarch-07 - 15 mins
[20:47:27] --- yangwoo has joined
[20:47:31] <dthaler> MANET Arch <http://www3.ietf.org/proceedings/07dec/slides/autoconf-1.ppt>
[20:51:21] --- kakima has left
[20:51:33] --- kakima has joined
[20:53:36] <dthaler> next is problem statement doc, clausen saying need to agree on what the doc should say (more than what it does say)
[20:55:22] <dthaler> now on problem statement document, but no slides online
[20:57:13] --- tclausen has joined
[20:59:23] <tclausen> (Dave Thaler on the mike, I fill in on jabber)
[20:59:47] <tclausen> Dave Thaler: what he thinks it should say
[21:00:01] <tclausen> DT: A MANET router that has hosts attached to it must be able to acquire an unique prefix.
[21:00:48] <tclausen> DT: A MANET router that does not have hosts attached must be able to get an unique address to configure its MANET interface.
[21:01:15] <tclausen> DT: In other words, it needs a space from which it can assign addresses to "things that are attached to it", and an address to assign to the MANET interface.
[21:01:40] <tclausen> DT: IF a prefix is on-link on another interface than the MANET, then one can not use that on the MANET interface.
[21:01:48] <tclausen> DT: So, these are two different things....
[21:02:04] <tclausen> DT: What you put on the non-MANET interfaces should, and I completely agree, an unique prefix.
[21:02:21] <tclausen> DT: Which is conditional on having hosts attached.
[21:02:32] <tclausen> Ryuji Wakikawa (RW): I agree.
[21:02:47] <tclausen> RW: A MANET router needs an address and a prefix. address not necessarilly in the ame range of the prefix.
[21:02:52] <dthaler> <thanks thomas :>
[21:02:57] <tclausen> RW: It's not clear from the problem statement.
[21:03:42] <tclausen> Joe Macker (JM): Not going to add to what Dave said, but want to confirm that this is important.
[21:03:54] <tclausen> <Dave, I'll hand over to you>
[21:03:56] <dthaler> macker: I have a case where it's just a router, no host there, so the conditional is impotant
[21:06:53] <dthaler> two cases... subordinate is where attached to outside world and get address space from there; and autonomous where not attached and MANET can choose space itself
[21:07:11] <dthaler> (who is at mike?)
[21:07:24] <tclausen> Alexandru Petrescu
[21:07:25] <TJ> alexandru petrescu
[21:07:29] <tclausen> (if I spell it correctly :) )
[21:07:30] <dthaler> thx
[21:07:49] <dthaler> AP: what is a "unique prefix", is there one per external network?
[21:08:15] <dthaler> i think he's asking whether the MANET can be multihomed?
[21:10:14] <dthaler> ?: if you just have 1 host can it use the manet address
[21:10:28] <dthaler> DT: no, manet arch says host can't see manet interface so can't use that address
[21:11:09] <dthaler> ?: what's a MANET local address? (note this term is not on the slide, question is apparently about the doc)
[21:11:32] <TJ> This is Ryuji Wakikawa from KEIO university at the mic.
[21:13:34] <dthaler> TC: we need to define how a router can get a prefix if it has stuff hanging off it
[21:14:26] <dthaler> TC: if router has a host portion, it's just like having any other host hanging off the router
[21:15:00] --- yangwoo has left: Computer went to sleep
[21:15:18] <dthaler> TC: if cloud is attached to outside, then need space from outside in want to communicate with outside
[21:15:27] <dthaler> TC: can be multihomed
[21:15:41] --- florent.parent@gmail.com has joined
[21:15:57] --- yangwoo has joined
[21:15:57] <dthaler> AP: if multihomed then not "unique" prefix but multi-prefixed
[21:17:48] --- Will Ivancic has left: Computer went to sleep
[21:18:39] <dthaler> (very confusing discussion going on, don't know if anyone is following AP)
[21:20:04] <yangwoo> If prefixes are to be delegated to other MRs, then prefixes are not unique anymore?
[21:21:03] <dthaler> answer is no they're still unique in that there's assigned to exactly one non-MANET link behind a MR
[21:21:12] <dthaler> s/there's/they're/
[21:21:31] <yangwoo> Then I failed to followup AP. :-(
[21:21:40] <dthaler> me too
[21:23:51] <dthaler> ron: if all you have is just a single host then should the router drop the packet
[21:24:04] <dthaler> TC: if you have a host attached even if colocated then it needs a prefix
[21:24:30] <dthaler> (this is what the manetarch doc says)
[21:26:10] <tclausen> DThaler: the reason that hosts even if only a single colocated host on a manet router, do not get access to the manet interface is, that the manet interface is so different from other interfaces
[21:26:20] <TJ> Dave Thaler: The current status according to the MANET arch doc: You get semantics that the normal IP model gives today. So the hosts and applications don't get what they expect. So MANET says, we're not going to worry about the impact on applications. So the host sees the ... that it's had all along
[21:26:44] --- heer.cs has left
[21:26:45] <tclausen> DThaler: hence, the manet architecture document insulates the MANET behavior from hosts.
[21:27:01] <TJ> Charles Perkins: You can have useful MANET devices that don't have these prefixes allocated to them, and they work just fine.
[21:27:18] <TJ> CP: It takes a little morework to correctly handle a subnet prefix and make sure it's reachable
[21:27:45] <TJ> in many circumstances, this extra work would just go to waste, so I'm voting for relaxing the requirement, but you should enable it in a consistent way
[21:28:20] <dthaler> side comment (not as scribe) CP thinks it should be okay to present different semantics to hosts (this is not current wg consensus)
[21:28:33] <TJ> CP: We had exactly the same discussion a year ago, and this architecture was drawn up that gives different semantics
[21:32:07] --- TJ has left: Computer went to sleep
[21:34:34] --- xiaohunhun has joined
[21:36:19] <tclausen> DThaler: wants to talk about if one can be multi-homed to two non-stubs, i.e. two different address spaces
[21:36:42] <tclausen> DT: natural assumption: it'd be allowed but not required since that is how it is for other networks.
[21:37:09] <tclausen> DT: especially since there's mobility etc., and one should allow nodes to get addresses/prefixes from multiple providers.
[21:37:26] <tclausen> DT: the statement should say if this is something that is prohibited or not.
[21:37:42] <tclausen> DT: if allowed, then coherent with usual site-multi-homing.
[21:37:59] <tclausen> DT: unique prefixes -- each router gets something that no other router has, but they may get "multiple of those"
[21:38:43] <dthaler> well, coherent with _one_ type of site multihoming anyway (not _the_ way to do site multihoming)
[21:38:53] <tclausen> (sorry, yes, was trying to type fast ;) )
[21:39:10] <intvelt> Justin Dean at the mike
[21:40:43] <dthaler> gabriel montenegro (GM): is it worth documenting what special applications get
[21:42:50] <dthaler> DT: yes it is. manetarch sort of does that now, calls them MANET-capable or MANET-aware protocols/apps or some such term. MANET routing protocols are an example of them.
[21:43:35] <dthaler> DT: maybe there's things missing, not sure but it does talk about some attributes/differences they see
[21:45:07] <dthaler> JM: still not clear what group wants to solve. is it just addressing?
[21:45:22] <dthaler> JM: e.g. gateway discovery?
[21:46:42] <dthaler> TC: currently just address assignment problem
[21:48:44] <dthaler> AP: we've only talked about goals... what protocols will we work on
[21:48:54] <dthaler> AP: should movement scenarios be listed
[21:50:54] <dthaler> AP: when we will know what protocols we'll work on?
[21:51:14] <dthaler> presenter: when the WG closes ;-)
[21:52:07] --- Will Ivancic has joined
[21:52:54] <dthaler> TC: we need to quickly finish agreeing on what the problem is, and THEN move on to the solutions as soon as we can, but need to focus on agreeing on the problem first since we have a hard time doing that much
[21:53:55] <dthaler> TC: movement scenario is in a doc already
[21:55:41] <dthaler> ?#3: would like to see examples of why existing protocols don't work
[21:55:54] <tclausen> ?#3 == Ulrich Herberg
[21:56:42] <dthaler> RW: agree.
[21:57:24] <dthaler> RW: add individual section for each major existing solution showing how it doesn't solve the problem
[21:58:41] <dthaler> JM: add multihome specifics, say it's a scenario, since it changes the solution space
[21:58:58] <dthaler> presenter: there is a multihoming section already
[21:59:22] <dthaler> JM: wasn't obvious to me from the diagrams, so needs to be clearer
[21:59:53] <dthaler> (presenter = Emmanuel Baccelli)
[22:02:30] --- john.zhao has joined
[22:03:16] --- teemu has joined
[22:05:01] <dthaler> UH: scope of addresses is unclear, maybe add a section on address scope?
[22:08:05] <dthaler> TC: would like to have next version have clearer description of the 2 connectivity scenarios that were presented,
[22:08:19] <dthaler> TC: and when does a router need to acquire a prefix, when does it need to acquire an address,
[22:08:58] <dthaler> TC: try to define what we mean by unique prefix, multihoming,
[22:09:55] <dthaler> TC: transit network scenario, say what we're trying to do (just addressing)
[22:10:22] <dthaler> (and many of these are already in the doc, just summarizing what we want to make sure is covered in the final doc that has been discussed today)
[22:11:07] <dthaler> TC: would like stuff posted to the list this week if possible
[22:11:45] <dthaler> TC: update draft within 2 weeks, eventual goal is to get it to the IESG prior to next IETF.
[22:13:21] <dthaler> JM: not clear whether a solution has to meet all the goals or just some of them
[22:14:36] <dthaler> JM: for example, is dealing with merging a hard requirement?
[22:14:57] <dthaler> JM: or can a solution proposal just say which ones it meets and which it doesn't?
[22:15:18] <dthaler> TC (personal opinion): would like a solution that meets all of them
[22:15:47] <dthaler> JM: great if can do all with low complexity, but if can do a subset with much less complexity then may be worth considering
[22:17:08] <dthaler> next
[22:17:09] <dthaler> o Ad-Hoc IP Autoconfiguration Solution Space Analysis draft-bernardos-autoconf-solution-space-00 - 10 mins
[22:17:24] <dthaler> Evaluation <http://www3.ietf.org/proceedings/07dec/slides/autoconf-2.pdf>
[22:19:23] <dthaler> wrong slides, should be Solution <http://www3.ietf.org/proceedings/07dec/slides/autoconf-4.pdf>
[22:24:42] --- igarashi has joined
[22:24:52] --- teemu has left
[22:25:42] <dthaler> TC asked who's read this document
[22:25:51] <dthaler> some people have (just an individual submission at present)
[22:26:20] <dthaler> but not very many, encourage people to read it
[22:27:06] <dthaler> next
[22:27:06] <dthaler> o Evaluation Considerations for IP Autoconfiguration Mechanisms in MANETs draft-bernardos-autoconf-evaluation-considerations-01 - 15 mins
[22:27:31] <dthaler> this is Evaluation <http://www3.ietf.org/proceedings/07dec/slides/autoconf-2.pdf>
[22:28:56] --- heer.cs has joined
[22:32:40] --- shubranshu has joined
[22:37:52] --- igarashi has left
[22:39:01] <dthaler> SS: running short on time, so please send comments on mailing list
[22:39:24] <dthaler> next
[22:39:24] <dthaler> o Requirements for IP Autoconfiguration Mechanisms in Backbone Wireless Mesh Network scenarios draft-bernardos-autoconf-backbone-mesh-reqs-00 - 15 mins
[22:39:32] <dthaler> Mesh Scenario <http://www3.ietf.org/proceedings/07dec/slides/autoconf-3.pdf>
[22:43:18] <dthaler> TC: send comments to list
[22:43:30] <dthaler> next
[22:44:02] <dthaler> o NEMO for Mobile Ad-hoc Networks draft-boot-autoconf-nemo4manet-00.txt - 15 mins
[22:44:14] <dthaler> I don't see the slides at https://datatracker.ietf.org/meeting/70/materials.html
[22:44:57] <dthaler> if anyone can find a URL for these slides, please post it here
[22:45:45] <dthaler> BR information flooded across MANET
[22:46:40] <shubranshu> ongoing presentation slides are available at https://datatracker.ietf.org/meeting/70/materials.html
[22:46:55] --- florent.parent@gmail.com has left
[22:47:08] <dthaler> MR's then select BR(s)
[22:47:16] <dthaler> (ok the slides showed up, I had to refresh)
[22:47:21] <dthaler> Nemo4MANET <http://www3.ietf.org/proceedings/07dec/slides/autoconf-6.pdf>
[22:47:47] <dthaler> generate address from BR prefix and own interface ID
[22:49:13] --- xiaohunhun has left: Replaced by new connection
[22:49:18] --- john.zhao has left: Replaced by new connection
[22:49:18] --- john.zhao has joined
[22:49:22] --- teco has left: Replaced by new connection
[22:49:52] <dthaler> use Mobile IP (but I didn't follow what for)
[22:50:34] <dthaler> ok, Mobile IP used to move to another BR (in case of partition? or to optimize return path?)
[22:51:01] <dthaler> uses NEMO
[22:51:34] --- kakima has left
[22:51:37] --- Will Ivancic has left: Computer went to sleep
[22:52:12] <dthaler> TC: send comments to list
[22:52:27] --- shubranshu has left
[22:53:43] --- intvelt has left
[22:54:27] --- tclausen has left: Logged out
[22:55:21] --- john.zhao has left: Computer went to sleep
[22:55:45] --- tom5760@gmail.com has left
[22:57:51] --- heer.cs has left
[22:59:55] <dthaler> problem statement slides (discussed near the top of this transcript) are at Nemo4MANET <http://www3.ietf.org/proceedings/07dec/slides/autoconf-6.pdf>
[23:07:12] --- yangwoo has left: Disconnected.
[23:24:39] --- dthaler has left