[01:54:09] --- thomas.heide.clausen@gmail.com has joined
[01:56:20] --- thomas.heide.clausen has joined
[01:56:50] --- thomas.heide.clausen has left
[01:56:58] --- thomas.heide.clausen@gmail.com has left
[08:49:06] --- jeffa has joined
[08:57:12] --- thomas.heide.clausen@gmail.com has joined
[08:59:18] --- narten has joined
[09:01:12] --- dthaler has joined
[09:04:09] <jeffa> calls for jabber scribe, I volunteered
[09:04:29] <jeffa> Thomas: we want to have lots of good discussion
[09:04:43] <jeffa> http://www3.ietf.org/proceedings/06jul/agenda/autoconf.txt
[09:04:51] <jeffa> agenda bashing, document status update
[09:05:26] <jeffa> Joe Macker will talk about the arch doc
[09:05:38] --- idc has joined
[09:05:49] <jeffa> Emmanuel will talk about ...
[09:05:57] --- rdroms@jabber.org has joined
[09:06:22] <jeffa> ...document has been upload to the autoconf blog
[09:06:39] <jeffa> there will be a presentation on common framework for autoconf
[09:06:41] --- irino has joined
[09:06:49] <jeffa> GAP is a new proposal to autoconf wG
[09:07:05] <jeffa> we will hear an update on how the protocol works, try and get some feedback
[09:07:22] <jeffa> multi gateway draft, then autoconf using DHCP
[09:07:38] <thomas.heide.clausen@gmail.com> Throwaway comment: Slides have now been received from all presenters, and I am uploading them to the usual place (https://datatracker.ietf.org/public/meeting_materials.cgi?meeting_num=66)
[09:07:42] --- yangwoo ko has joined
[09:07:59] <jeffa> request for comments
[09:08:02] <jeffa> on the agenda
[09:08:07] <jeffa> (no comments)
[09:08:57] --- badamson has joined
[09:08:57] --- badamson has left: Lost connection
[09:09:12] <jeffa> Joe Macker is presenting ... he will keep it "short and sweet"
[09:09:14] --- narten has left
[09:09:20] <jeffa> he wants to give a status on what they are doing
[09:09:25] <jeffa> and hopefully get some comment
[09:09:29] <jeffa> just two pages to talk about
[09:09:45] <jeffa> slides: MANET architecture ID
[09:09:47] --- jgre has joined
[09:10:18] --- rjaksa has joined
[09:10:21] --- badamson has joined
[09:10:21] --- badamson has left: Lost connection
[09:10:31] --- narten has joined
[09:10:35] <jeffa> MANET WG also gives input
[09:10:42] <jeffa> publish -00 after IETF
[09:10:46] <jeffa> slide : ID Progress
[09:11:07] <jeffa> they want to keep design open to this group
[09:11:17] <jeffa> Ian Chakeres is the lead edtior
[09:11:21] <jeffa> send comments to him or the ML
[09:11:37] <jeffa> Joe: ** please comment ** on this document
[09:11:49] <jeffa> they want to keep it short, publish it as informational
[09:12:07] --- narten has left
[09:12:16] <jeffa> not a completely finished outline... main thing is discuss challenges and disgn considerations
[09:12:40] <jeffa> it starts discussing taxonomies
[09:12:50] --- narten has joined
[09:12:59] <jeffa> also creates scenario/deployment examples
[09:13:17] --- badamson has joined
[09:13:18] <jeffa> will try to get this doc out within two weeks
[09:13:21] <jeffa> so people can comment
[09:13:37] --- badamson has left: Lost connection
[09:13:39] <jeffa> Joe is going over the outline. slide: Draft Outline Developed
[09:13:43] --- badamson has joined
[09:13:43] --- badamson has left: Lost connection
[09:15:04] <jeffa> discussing types of taxonomies
[09:15:26] <jeffa> discussing the number of nodes is not adequate
[09:16:18] <jeffa> or is not perfect; traffic/mobility models, heterogeneity...
[09:16:39] <jeffa> concentrate on IP issues, not just one wireless interface type
[09:17:46] <jeffa> not sure about IANA/Security sections, this is informational
[09:18:01] <jeffa> (each protocol needs to address its own)
[09:18:11] <jeffa> ask questions now
[09:18:13] <jeffa> dave thaler:
[09:18:15] --- badamson has joined
[09:18:20] --- badamson has left: Lost connection
[09:18:27] <jeffa> DT: one thing I don't see is the discussion of the subnet model
[09:18:43] <jeffa> has written something before ... will it be incorporated
[09:18:51] <jeffa> Joe: I think we will incorporate the basic concept
[09:18:58] <jeffa> we don't want to get into addressing schemese
[09:19:12] <jeffa> which is impl dependant
[09:19:37] <jeffa> there will be a section about MANET routers could include some of this information
[09:19:48] <jeffa> DT: There are some parts that are part of a solution
[09:19:55] <jeffa> but there is a core architectural issue here
[09:19:58] <thomas.heide.clausen@gmail.com> The I-D Dave is talking about is: http://manetautoconf.online.fr/Blog/wp-content/draft-thaler-intarea-multilink-subnet-issues-00.txt
[09:19:58] <jeffa> that would be appropriate
[09:20:08] <jeffa> Joe: appreciate the comment
[09:20:23] <jeffa> Jari: Daves concerns, w/ a slightly different angle
[09:20:37] <jeffa> you are presenting issues, ways to configure MANET networks, it would be more useful to
[09:20:42] <jeffa> nail down the architectural model
[09:20:53] <jeffa> b/c it's hard to design the protocols we need if all we have is
[09:21:04] <jeffa> a set of examples
[09:21:12] <jeffa> Joe: is there a one size fits all solution?
[09:21:27] <jeffa> Jari: not sure there is a one size fits all
[09:21:34] <jeffa> Joe: this is a living document
[09:21:43] <jeffa> designers can have some debate on
[09:22:01] <jeffa> hopefully it will be useful, otherwise let's change it
[09:22:21] <jeffa> there is some need for some tutorialist, historical type information
[09:22:39] <jeffa> some papers have claimed things that may not be true in IETF MANET
[09:23:02] <jeffa> we need to have lots of discussion if this doesn't do its job
[09:23:08] --- badamson has joined
[09:23:10] <jeffa> DT: one arch or multiple architectures?
[09:23:24] <jeffa> the addressing model is actually an arch. and not a solution
[09:23:34] <jeffa> a solution is a specific protocol, so to me the subnet level
[09:23:37] <jeffa> is the architectur
[09:23:47] <jeffa> Joe: ...
[09:23:59] <jeffa> Jari : it would be useful to have at least one architecture initially
[09:24:10] <jeffa> let's pick one and go forward
[09:24:13] <jeffa> Fred Templin:
[09:24:27] <jeffa> I don't want to presuppose, but define what the term "link" means
[09:24:30] <jeffa> Charles Perkins:
[09:24:33] <jeffa> curious ...
[09:24:39] <jeffa> question of subnets seems quite contentious
[09:24:53] <jeffa> there is a lot of important information that needs to be put into a document
[09:25:08] <jeffa> your opinion: how much of the document will be taken up
[09:25:15] <jeffa> trying to define subnet models?
[09:25:40] <jeffa> Joe: subnets are an issue, < talking about subnets >, I am more of a fan of the classical IP model
[09:26:01] <jeffa> not a fan of not allowing interesting stuff to happen... more into what are links, what does that mean
[09:26:11] <jeffa> CP: it is worthwhile to keep in mind we went along
[09:26:20] <jeffa> in MANET WG for many years without really having subnets
[09:26:27] <jeffa> Joe: it's an Ethernet type thing... ...
[09:26:33] <jeffa> CP: let's get something useful done
[09:26:44] <jeffa> even if we can't define subnets to everybody's satisfaction
[09:26:56] <jeffa> Joe: That may be one of the points people need to be very
[09:26:58] <jeffa> careful about
[09:27:14] <jeffa> we could do this correctly and keep this simple enough
[09:27:24] <jeffa> this is not supposed to solve all the design challenges
[09:27:30] <jeffa> but provide a framework
[09:27:33] --- petrescu7 has joined
[09:27:35] <jeffa> for dicussion
[09:27:37] <jeffa> Thomas:
[09:28:08] <jeffa> subsequent to this presentation there will be a discussion on the problem statement.... can you say some words about the aim of the descriptions that will be in the arch document
[09:28:16] <jeffa> Joe : restate the question...
[09:28:29] <jeffa> The scope should be fairly simple
[09:28:32] <jeffa> of the arch doc
[09:28:42] <jeffa> what concept may stimulate various autoconf issues
[09:28:54] <jeffa> for example ...
[09:29:18] <jeffa> nodes using multi-hop DHCP, fragmenting nodes, autonomous operation
[09:29:28] <jeffa> one of the key things is to talk about fixed infrastructure connection
[09:29:42] <jeffa> what is your intent for the applications?
[09:29:52] <jeffa> do you really want these things to be p2p?
[09:30:08] <jeffa> the autoconf solutions in one case are much simpler than in another case
[09:30:17] <jeffa> no different than internet today -- dhcp, zeroconf, etc.
[09:30:27] <jeffa> the point is not to make the design decision
[09:30:34] <jeffa> but to provide for dicussion, rationale
[09:30:52] <jeffa> Thomas: there is a draft about connectivity scenarios
[09:31:07] <jeffa> also you can listen to the audio from previous meetings
[09:31:14] <jeffa> Next up is Emmanuel.
[09:31:20] <thomas.heide.clausen@gmail.com> Problem Statement Slides: http://www3.ietf.org/proceedings/06jul/slides/autoconf-6.ppt
[09:31:42] <jeffa> < switching computers >
[09:31:52] <thomas.heide.clausen@gmail.com> <sorry 'bout that>
[09:31:59] <jeffa> MANET autoconf problem statement
[09:31:59] --- kazuya has joined
[09:32:01] <rjaksa> who is the speaker?
[09:32:04] <jeffa> Slide: status
[09:32:14] <thomas.heide.clausen@gmail.com> Emmanuel Bacceli is the speaker
[09:32:25] <jeffa> converging ADP, framework, and scenario draft --> single doc
[09:33:02] <jeffa> Slide: Contents
[09:33:09] <jeffa> going over the table of contents
[09:33:43] <jeffa> slide: Constraints
[09:34:04] <jeffa> talking about different contrsatins
[09:34:08] <jeffa> constraints
[09:34:35] <jeffa> dynamic topoology, possible standalone, possibly merging
[09:34:44] <jeffa> and the different security concerns
[09:34:51] <jeffa> slide: considerd scenarios
[09:35:13] <jeffa> looking at standalone MANET, connected (stub), and multiple gateways
[09:35:17] <jeffa> Joe Macker:
[09:35:25] <jeffa> let's use a different term than "stub" networks
[09:36:08] --- irino has left: Lost connection
[09:36:08] --- petrescu7 has left: Lost connection
[09:36:08] --- kazuya has left: Lost connection
[09:36:08] --- jeffa has left: Lost connection
[09:36:08] --- rjaksa has left: Lost connection
[09:36:08] --- rdroms@jabber.org has left: Lost connection
[09:36:08] --- dthaler has left: Lost connection
[09:36:08] --- jgre has left: Lost connection
[09:41:48] --- jeffa has joined
[09:42:08] <jeffa> sorry ... unable to connect to jabber.org
[09:42:22] <jeffa> Joe Macker: is DAD always useful
[09:42:38] <jeffa> previously -- some comments by Charlie Perkins that I missed
[09:42:59] <jeffa> please don't make it (DAD?) a requirement
[09:43:09] <jeffa> EB: pre-service DAD
[09:43:37] <jeffa> JM: don't want to require this for the protocol
[09:44:32] <jeffa> dad is not something everybody needs
[09:44:43] <jeffa> leave a place in the state machine that you can turn it off
[09:45:02] <jeffa> EB: agree that we need to be clearer in the document
[09:45:10] <jeffa> JM: I would like DAD to be optional.
[09:45:13] <jeffa> able to turn it off
[09:45:26] --- narten has left
[09:45:37] <jeffa> Dave Thaler: we haven't seen the actual draft yet, it's fuzzy to me what is the boundary between two documents...
[09:45:48] <jeffa> the state machine is a specific solution
[09:45:54] <jeffa> it is not the only way to do that
[09:46:06] <jeffa> if this is a problem statement, the state machine shouldn't be there
[09:46:10] --- momose has joined
[09:46:23] <jeffa> this is one way to solve the problem
[09:46:30] <jeffa> JB: ok, I see your point.
[09:46:43] <jeffa> try to determine different modules that are needed
[09:46:56] <jeffa> goal of the state machine is to determine what kind of components are needed
[09:47:15] <jeffa> DT: ok, but notion of tentative address is not always there for example
[09:47:17] --- irino has joined
[09:47:23] <jeffa> shouldn't be part of the problem statement
[09:47:45] <jeffa> it's probably ok to talk about these things, as long as you clearly say "for example"
[09:48:11] <jeffa> EB: ...
[09:48:39] <jeffa> DT: < talking about different solutions, dividing up the address space >
[09:48:56] <jeffa> Thomas: I don't disagree with what's been said
[09:48:58] --- narten has joined
[09:49:07] --- rjaksa has joined
[09:49:07] <jeffa> the way I look at the state machine is more of a general model
[09:49:18] <jeffa> saying you may need to take this or that transition
[09:49:33] <jeffa> be careful when we describe this so that we are not mandating these things
[09:49:56] <jeffa> think of this more of a model of how to do things instead of a solution
[09:50:13] <jeffa> we should try to describe it within that context
[09:50:55] <jeffa> Chris D: it's important that this looks like a solution but nodes may have an address but not a root (?)
[09:51:20] <jeffa> (unk?) : in my opinion this should not be a requirement but more of a reference model
[09:51:33] <jeffa> solution does not always require a reference model
[09:51:40] <jeffa> Joe Macker:
[09:51:52] <thomas.heide.clausen@gmail.com> (The person referenced as "unk" was Kenichi Mase....)
[09:52:03] <jeffa> when I look at this, I know it helps me understand this... but there are missing arrows here
[09:52:10] <jeffa> I am concerned with having this in the problem statement.
[09:52:13] <jeffa> EB: Ok.
[09:52:27] <jeffa> THC: 2nd the comment from Dave Thaler from before...
[09:52:45] <jeffa> synchronize the content and scope between those two documents
[09:52:52] <jeffa> rather than do this the week before the IETF
[09:53:09] <jeffa> Emmanuel -- returning to the presentation -- slide : Framework
[09:53:35] --- dthaler has joined
[09:53:55] <jeffa> slide: Goals and Requirements
[09:54:16] <jeffa> THC: the message in this slide is not that the goal of this document is to provide solutions
[09:54:40] <jeffa> Joe Macker: yeah, I'm continuing to see solutions. Problem statements are rationale for solutions, that is what we should do.
[09:54:50] <jeffa> make it very clear that it is rationale, not a solution
[09:55:07] <jeffa> EB: Yes, we are not talking about the algorithms, but designing the concepts needed
[09:55:13] <jeffa> slide: Where are we at
[09:55:39] <jeffa> we are working in parallel with the arch draft
[09:55:55] <jeffa> we don't have initial text about security considerations
[09:56:12] <jeffa> check out these links : http://manetautoconf.online.fr/Blog
[09:56:17] <jeffa> and the tracker
[09:56:42] <jeffa> THC: The message here is that the objective is to try and get this out
[09:56:55] <jeffa> and we need a thorough read-through of this and the other document
[09:57:26] <jeffa> specifically, are the scenarios you are thinking about permitted and falling into the framework?
[09:58:22] <jeffa> Kenichi Mase draft-mase-autoconf-gap-00
[09:58:32] <jeffa> oops
[09:58:40] <jeffa> draft-mase-autoconf-framework-02
[09:58:52] <jeffa> slide: Aims
[09:59:21] <jeffa> (topic is "A Common Framework for Autoconfiguration")
[09:59:33] <dthaler> http://www3.ietf.org/proceedings/06jul/slides/autoconf-0.ppt (I'm in another wg now and following via jabber)
[09:59:34] <jeffa> changes from 01
[09:59:36] <thomas.heide.clausen@gmail.com> Framework slides: http://www3.ietf.org/proceedings/06jul/slides/autoconf-0.ppt -- presenter: Kenichi Mase
[09:59:49] <jeffa> slide: a concept of DAA
[10:00:16] <jeffa> discussing the aims of DAA
[10:00:48] <jeffa> slide: baselines of the framework
[10:01:02] <dthaler> what subnet model does this framework assume, or is it independent of the subnet model? (the slides only say "address" singular)
[10:01:11] <jeffa> discussing the behavior of the nodes
[10:01:33] --- wrstuden has joined
[10:01:35] <jeffa> the benefit of this is to reduce overhead
[10:01:39] <jeffa> question from Joe Macker
[10:02:09] <rjaksa> reduce what kind of overhead?
[10:02:10] <jeffa> JM: idea from Joe Macker ... (I missed it)
[10:02:41] <jeffa> Fred Templin: going back to what Joe said ... I think these are necessary mechanisms with link local addresses, but with DHCP you get global addresses delegated by an authority
[10:03:03] <jeffa> THC: we need to be careful in this document... not saying each node should perform
[10:03:16] <jeffa> but the AUTOCONF suite should provide this mechanism if needed
[10:03:19] --- kazuya has joined
[10:03:28] <jeffa> be clear in the wording that all nodes must not do these things
[10:03:38] <jeffa> Robert Jackson: what kind of overhead does this reduce?
[10:04:07] <jeffa> Kenichi: when we use routing messages, we can reduce overhead
[10:04:11] <jeffa> for example OLSR
[10:04:24] <jeffa> we can use OLSR messages
[10:04:45] <jeffa> there is a general packet/message format
[10:04:58] <jeffa> so overhead is almost zero (if we use this DAD with it)
[10:05:09] <jeffa> Joe Macker: some people have done this using OLSR
[10:05:19] <jeffa> the issue is that this is an extension to a routing protocol
[10:05:27] <jeffa> you have to assume that it gets everywhere
[10:05:37] <jeffa> SMF provides something that is protocol independent
[10:05:50] <jeffa> as a framework we have to be careful to not force a solution
[10:05:54] <jeffa> Kenichi: it's a reference model
[10:06:02] <jeffa> Joe: yes but it says how to do it
[10:06:14] <jeffa> we want to talk about those (?) as alternatives
[10:06:39] <jeffa> Kenichi: people want DAD to be optional because of lots of overhead
[10:06:47] <jeffa> here overhead is almost zero, so why not?
[10:07:06] <jeffa> THC: it is an interesting document, but I 2nd the desire to see it separated /abstracted from an actual solution
[10:07:13] --- momose has left
[10:07:22] <jeffa> for example integrating this into OLSRv2 is a good idea, but this WG doesn't want to reference
[10:07:26] <jeffa> once protocol or the other
[10:07:37] <jeffa> and there are MANETs with no routing protocol running but they need this
[10:07:50] <jeffa> takeway -- let's try and make this more genereal instead of a specific solution
[10:08:04] <jeffa> there is no standards track MANET protocol
[10:08:28] <jeffa> further comments?
[10:08:47] <thomas.heide.clausen@gmail.com> Gateway Aggregation Protocol slides: http://www3.ietf.org/proceedings/06jul/slides/autoconf-1.ppt -- presenter: Kenichi Mase
[10:09:00] <jeffa> --------- switching to next presentation ------- draft-mase-autoconf-gap-00
[10:09:19] <jeffa> slide: problem statement and motivation
[10:09:55] <jeffa> goal is to avoid change in the global address, so we can skip DAD
[10:10:03] <jeffa> since the address does not change
[10:10:13] <jeffa> need to inform other nodes
[10:10:35] <jeffa> to realize this approach, we use the same prefix from different MNs
[10:10:44] <jeffa> slide: GAP mechanism
[10:10:48] <jeffa> (diagram)
[10:11:00] <jeffa> describing the diagram
[10:11:11] <dthaler> what subnet model does this solution assume, or is it independent of the subnet model? (the slides only say "address" singular)
[10:11:17] <jeffa> edge = they have another interface w/Internet connencted
[10:12:36] --- wrstuden has left: Computer went to sleep
[10:12:39] <jeffa> describing the scenario
[10:12:59] <jeffa> Thomas -- can you ask Dave's question using the mic?
[10:13:07] <thomas.heide.clausen@gmail.com> yes....
[10:13:35] <jeffa> THC: stating Dave's question
[10:14:09] <jeffa> Kenichi: he is thinking this is the same subnet
[10:14:16] <thomas.heide.clausen@gmail.com> Same subnet - Dave, does this help?
[10:14:17] <jeffa> (pointing to the diagram)
[10:14:27] <dthaler> well it answers my question :)
[10:14:36] <thomas.heide.clausen@gmail.com> dthaler: ok, good.
[10:14:43] <jeffa> < I guess it's a single subnet model >
[10:14:49] <dthaler> whether it helps or hinders is another matter.
[10:14:59] --- jariarkko has joined
[10:15:01] <thomas.heide.clausen@gmail.com> dthaler: ;)
[10:15:18] <jeffa> slide: A Procedure of ...
[10:15:34] <jeffa> describing what happens when an MN becomes an Edge MN
[10:15:40] <jeffa> (following diagram left to right)
[10:16:01] <jeffa> questions/comments?
[10:16:36] <jeffa> : you use terminology which is different than the current terminology we use in the problem statement
[10:16:53] <jeffa> the role that you give to the Edge MN, is it right or not?
[10:17:12] <jeffa> your Edge MN == Internet Gateway in the problem statement
[10:17:52] <jeffa> (question was from S. Ruffino)
[10:18:00] <jeffa> does the AR run MANET routing protocol or not?
[10:18:12] <jeffa> Kenichi: no
[10:18:23] <rjaksa> the slides presented seem slightly different than the slides available.
[10:18:29] <jeffa> SR: a little confused about the solution
[10:18:52] <jeffa> seems complicated, and not trivial to handle all the signalling for the tunnels. that is my first impression. will let you know on the list
[10:18:57] <thomas.heide.clausen@gmail.com> rjaksa: that may be so, I put the last on-line which the authors gave me. I will ask for the slides. thanks for bringing it to my attention,
[10:18:59] <jeffa> Kenichi: I need to update this
[10:19:15] <jeffa> Joe Macker: I see cellular link, this may be a little different use case
[10:19:30] <jeffa> you are doing something different that is particular to a certain deployment scenario
[10:20:01] <jeffa> ... talking about adv mutliple prefixes ...
[10:20:15] <jeffa> you can subdivide the prefix further, it doesn't have to be one address
[10:20:30] <jeffa> my point was there may be other simpler architectures that may also need to be discussed
[10:20:44] <jeffa> how do you handle that other case (the simpler one?)
[10:20:57] <jeffa> this is different b/c now you have additional access link
[10:21:10] <jeffa> where would we put the other issue? problem statement/arch?
[10:21:21] <jeffa> the real discussion of multiple gateways, where does it belong?
[10:21:26] <jeffa> will have that discussion offline
[10:21:51] <jeffa> Kenichi: ...talking about his diagram, AR ... we need to discuss more about that
[10:22:08] <jeffa> could make the statement more general to accommodate this type of concept
[10:22:33] <jeffa> Ya...? : there is only one internet gateway on this slide, so why is it named gateway aggregation?
[10:23:03] <jeffa> Kenichi: here the number is one, traditionally edge MN is the gateway, that's why the term, since there are multiple edge MN
[10:23:19] <jeffa> SR:
[10:23:33] <jeffa> SR: single vs multiple prefix in a MANET
[10:23:40] <jeffa> when you have multiple gateways
[10:23:46] <jeffa> Kenichi: other questions?
[10:23:54] <jeffa> -------- switching presentations ---------
[10:24:08] <jeffa> draft-ruffino-manet-autoconf-multigw-03.txt
[10:24:13] <yangwoo ko> Ya....? was Yangwoo Ko
[10:24:28] <jeffa> (YK - thanks)
[10:24:39] <dthaler> http://www3.ietf.org/proceedings/06jul/slides/autoconf-4.pdf
[10:24:44] <jeffa> Automatic configuration of IPv6 ................. 20 min addresses for MANET with multiple
[10:25:07] <jeffa> Automatic configuration of IPv6 addresses for MANET with multiple gateways (AMG)
[10:25:10] <thomas.heide.clausen@gmail.com> Automatic configuration of IPv6 Gateway Addresses with Multiple Gateways slides: http://www3.ietf.org/proceedings/06jul/slides/autoconf-4.pdf -- presenter: Simone Ruffino
[10:25:17] <jeffa> we are constantly updating the draft
[10:25:19] <thomas.heide.clausen@gmail.com> Oops, Dave was ahead of me. Thanks Dave.
[10:25:30] <jeffa> slide: AMG overview
[10:26:41] <jeffa> using each gateway's metric, MANET nodes build a ranking
[10:26:44] <jeffa> slide: design goals
[10:26:55] <jeffa> talking about the design choices they made
[10:27:37] <jeffa> initially focused on dynamic scenarios, and where global addresses change
[10:28:14] <jeffa> Charles Perkins:
[10:28:36] <jeffa> in many cases, may be intermittent loss of connectivity
[10:29:33] <jeffa> SR: actually, the solution tried to provide a mechanism where you can use other gateways if they are available, with optimized operation
[10:29:38] <jeffa> in order to minimize the outage
[10:29:41] <dthaler> I'm assuming this solution presupposes a single-subnet model, correct?
[10:29:56] <dthaler> (one address per manet node, yes?)
[10:30:11] <jeffa> SR: we are working on the intermittent case
[10:30:35] <jeffa> SR: yes single subnet model
[10:30:47] <jeffa> continues presenting slide: Design goals
[10:31:41] <jeffa> slide: Design choices
[10:32:37] <jeffa> DAD is not specified ....
[10:33:22] <jeffa> Different gws adv different prefixes: use DHCPv6 w/prefix delegation option
[10:33:42] <jeffa> give example
[10:33:51] <jeffa> he is providing an example
[10:34:22] --- monden has joined
[10:34:38] <jeffa> slide: Design choices (cont)
[10:35:23] <jeffa> Fred Templin: why couldn't you also piggyback the prefix announcements on the routing messages?
[10:35:27] <jeffa> is SMF really needed?
[10:35:44] <jeffa> SR: the choice was made to make this independent from the routing protocol
[10:36:12] <jeffa> FT: there is a slight tradeoff in efficiency, but this may be negligible if you don't have a lot of traffic.
[10:36:14] <jeffa> SR: yes.
[10:36:22] <jeffa> slide: Phase 1: MANET ...
[10:36:37] <jeffa> the phases can happen concurrently
[10:36:48] <jeffa> but he is presenting them in order.
[10:36:54] <jeffa> slide: Phase 2: ...
[10:37:46] <jeffa> Joe Macker: to follow up on Fred's comment, since you are using packet bb and are defining TLVs, it should be fairly easy to jump back and forth between the different protocols
[10:37:52] <jeffa> since they use a common TLV format
[10:38:00] <jeffa> this approach can be flexible
[10:38:13] <jeffa> SR: ok, I understand. initially we used OLSRv1 message format
[10:38:19] <jeffa> I think it's easy to extend to OSLRv2
[10:38:48] <jeffa> THC: reinforce what Joe said, design this to the packet format, not to a specific routing protocol
[10:39:02] <jeffa> slide: Phase 3: ....
[10:39:59] <jeffa> proposes two algorithms, description of each
[10:41:04] <jeffa> slide: Phase 4 ...
[10:41:32] --- yowada has joined
[10:41:39] <jeffa> slide: AMG
[10:41:44] <jeffa> diagram w/animation
[10:41:54] <jeffa> gateways are C
[10:41:59] <jeffa> and other yellow laptops near top
[10:42:02] <jeffa> of cloud
[10:42:02] <thomas.heide.clausen@gmail.com> Notice: slides uploaded are in pdf, thus no animations....sorry, folks.
[10:42:48] <jeffa> Charlie Perkins: What happens if a node has an existing address, comes into the MANET with a gateway, and wants to configure another address for this MANET... so it has two addresses, is it going to be possible to do this
[10:42:54] <jeffa> without configuring the ULA also?
[10:43:00] <jeffa> SR: must think about it.
[10:43:19] <jeffa> we propose to use the ULA to handle the case where there is no gw from the begining
[10:43:28] <jeffa> you can use it, it should not cause
[10:43:36] <jeffa> any major problems as far as I could see
[10:43:50] <jeffa> I must reason about the different scope of addresses
[10:44:01] <jeffa> CP: the address may not be topologically correct
[10:44:08] <dthaler> Many IPv6 nodes try to configure multiple addresses by default... RFC 3041 (Privacy Extensions), so you shoudl assume (say) 9 addresses per host
[10:44:21] <jeffa> can't use it to receive packets from the net, but can use it for disconnected MANET
[10:44:24] <dthaler> (9=1 LL, 1 global public, 7 RFC3041)
[10:44:38] <jeffa> CP: we had an implementation of a gateway that did this...
[10:44:58] <jeffa> Fred Templin: it would always be prudent to use the MANET local address to operate the routing protocol
[10:45:07] <jeffa> as the node's router ID in the MANET
[10:45:29] <jeffa> CP: I guess I'll disagree, you can run the routing protocol using an address you already have
[10:45:33] <dthaler> the problem statement document should make it clear that one problem is to allow every host (note I use the word host here intentionally) to have as many addresses as it wants
[10:45:37] <jeffa> don't need to go through this process
[10:46:02] <jeffa> SR: the MANET local address can keep the protocol going also if the global addresses in use expire
[10:46:23] <jeffa> especially true for proactive routing protocols
[10:46:46] <jeffa> CP: ok, if a ULA is being suggested as a mechanism for overcoming address lifetime problems, then I am a little surprised
[10:46:57] <jeffa> why do you need to get an address if you already have one
[10:47:14] <jeffa> Fred Templin: chicken and egg ... link local address to participate
[10:47:37] <jeffa> CP: not true, example of using the internet, then disconnects and goes mobile to join a MANET
[10:47:46] <jeffa> FT: sure, that'd be ok
[10:47:59] <jeffa> SR: continues with animation
[10:48:23] <jeffa> Thomas -- we should address Dave's comments (see above)
[10:48:46] <thomas.heide.clausen@gmail.com> Argh, I missed that....sorry.
[10:48:49] <thomas.heide.clausen@gmail.com> Dave, want me to mike it?
[10:49:22] <jeffa> example: traffic will flow C-B-A
[10:49:52] <jeffa> John S.: I didn't understand the previous step
[10:50:00] <jeffa> in a MANET how do you know it's closest?
[10:50:06] <jeffa> SR: metric from the routing table?
[10:50:20] <jeffa> Joe Macker: logical hop, you are thinking of distance
[10:50:35] <jeffa> SR: I forgot to mention, the node can start the routing protocol concurrently
[10:50:51] <jeffa> after a while it learns the metric of the gateways that are around him
[10:51:03] <thomas.heide.clausen@gmail.com> John S: John Schnizlein....
[10:51:14] <jeffa> this is an incomplete example, only to highlight this situation
[10:51:35] <jeffa> situation is much more dynamic, default GW can change, find itself nearer to other GWs
[10:52:03] <jeffa> Joe Macker: important to remember "nearest", etc is just an example of a metric that you think is important
[10:52:10] <jeffa> we don't really know which metrics are important
[10:52:23] <jeffa> could flip-flop, ...
[10:52:39] <jeffa> SR: the best-hop selection does not mandate that the node change its gw
[10:52:54] <jeffa> there is a threshold so it does not change its ranking too often
[10:53:03] <jeffa> JM: you have a mechanism to control stickiness
[10:53:27] <jeffa> JM: do we know enough about how to set these parameters?
[10:53:39] <jeffa> SR: future work is to work on this parameter
[10:54:09] <jeffa> Chris D.: if you do want the shortest hop, you can use SMF floodest message since packet bb has a hopcount
[10:54:20] <jeffa> SR: is hopcount updated by nodes upon forwarding?
[10:54:24] <jeffa> CD: yes, it should be
[10:54:40] <jeffa> SR: not referring to TTL, but field within message
[10:54:42] <jeffa> CD: yes
[10:55:04] <jeffa> CD: also, I see an area where you may pick your nearest gw but you also may not pick your nearest
[10:55:11] <idc> my opinion about smf
[10:55:26] <jeffa> the routing protocol will naturally route packets towards the gw ...
[10:55:34] <jeffa> SR: let's talk offline
[10:55:36] <idc> smf does native IP multicast, therefore flooding via smf will not modify the contents of the packet
[10:55:42] <jeffa> THC: take this issue to the list
[10:56:01] <idc> therefore hopcount within packetbb will not be available
[10:56:02] <jeffa> slide: Best Prefix Selection
[10:56:41] <jeffa> slide: Global Adresses Advertising
[10:56:57] <jeffa> we introduced this for when nodes change their addresses...
[10:56:59] <thomas.heide.clausen@gmail.com> idc: may I ask that we take this to the M-L, rather than take it in the session?
[10:57:11] <idc> sure
[10:57:28] <thomas.heide.clausen@gmail.com> thanks - also thanks for your comment on this matter.
[10:57:38] <jeffa> (need a jabber2ml.pl script)
[10:57:44] <jeffa> slide: Pros and Cons
[10:58:09] <thomas.heide.clausen@gmail.com> jaffa: heh, yes ;)
[10:58:25] <jeffa> (the jabber logs are handy though)
[10:58:46] <jeffa> pros... also keep in mind what Charlie said before
[10:58:53] <jeffa> discussing cons
[10:59:08] --- julien.garnier@gmail.com has joined
[10:59:12] <thomas.heide.clausen@gmail.com> (yes, especially when compiling minutes afterwards)
[10:59:23] <jeffa> slide: Draft status
[10:59:31] <jeffa> currently -03 version
[11:00:01] <jeffa> http://vesuvio.ipv6.cselt.it/ruffino/
[11:00:10] <jeffa> check the link for updated versions of the draft
[11:00:48] <jeffa> THC: restating Dave Thaler's comment from the log
[11:00:57] <jeffa> comments?
[11:01:12] <jeffa> Charlies Perkins:
[11:01:27] <jeffa> to Dave Thaler -- I agree we should give out as many addresses as they want
[11:01:35] <jeffa> but that doesn't mean to give out addresses they don't need
[11:01:45] <jeffa> ---------- switching presentations -----------
[11:01:46] <jeffa> MANET Autoconfiguration using DHCP (draft-templin-autoconf-dhcp-01)
[11:01:53] <jeffa> speaker is Fred Templin
[11:01:58] <thomas.heide.clausen@gmail.com> Autoconfiguration using DHCP slides: http://www3.ietf.org/proceedings/06jul/slides/autoconf-2.ppt -- presenter: Fred Templin
[11:02:32] <jeffa> slide: MANET ... Problem Space
[11:02:39] <jeffa> (describing diagram)
[11:02:50] <jeffa> slide: .. Alternatives
[11:03:36] <jeffa> slide: Goals
[11:04:08] <jeffa> slide: ... Using DHCP
[11:04:28] <jeffa> as Charlie pointed out, they could also use a previously allocated address
[11:04:34] <thomas.heide.clausen@gmail.com> (I think that the slides on-line are different from those presented - in a different order. current slide is "MANET Autoconfiguration using DHCP"
[11:04:54] <jeffa> slide: Extended RS, RA, and DHCP Msgs
[11:05:51] <jeffa> diagram on messages
[11:06:25] <jeffa> esentially we are tunneling multicast within unicast or mcast in mcast
[11:06:30] <jeffa> slide: Model of Operation
[11:06:59] <jeffa> what is interesting is that we are tying together neighbor discovery and DHCP operation
[11:07:34] <jeffa> slide: Model of Operation (cont'd)
[11:08:19] <jeffa> slide: Router/Tunnel .... ... MG1
[11:08:28] <jeffa> diagram of this situation
[11:09:01] <jeffa> describes the configuration
[11:09:17] <jeffa> next slide -- (animation)
[11:09:42] <jeffa> next slide -- (animation) (same title)
[11:09:50] <jeffa> slide : ... MG2
[11:10:06] <jeffa> SR: quick question, what method do you use to let the MR1 choose the new gateway?
[11:10:28] <jeffa> Fred Templin: they hear advertisements from multiple gateways
[11:10:46] <jeffa> I see many similarities instead of differences in what is being briefed here.
[11:10:56] <jeffa> slide ...
[11:11:02] <jeffa> slide : additional Considerations
[11:12:48] <jeffa> SR: the problem I see is that most of the traffic will be downlink
[11:13:05] <jeffa> so actually, do you think that your mechanism acheives a good distribution of the downlink traffic ...
[11:13:12] <jeffa> in a way that optimizes the performance of the MANET nodes?
[11:13:28] <jeffa> dividing traffic among gateways may not be enough
[11:13:41] <jeffa> FT: each MANET router is going to make an independent decision
[11:13:53] <jeffa> maybe what's missing here is some coordination between MANET routers
[11:13:59] <jeffa> so that traffic is not concentrated
[11:14:02] <jeffa> SR: Ok, thanks.
[11:14:10] <jeffa> FT: continues...
[11:14:19] <jeffa> that is basically what goes on inside of the MANET
[11:14:24] <jeffa> ------ switching slides -----------
[11:14:33] <thomas.heide.clausen@gmail.com> (NETLMM - basically, goes on to the very first slide of the ppt that was uploaded....)
[11:14:39] <jeffa> netlmm mangament using DHCP -- Fred Templin
[11:15:00] <jeffa> summarizes
[11:15:10] <jeffa> this is now talking about what goes on in the backhaul network
[11:15:18] <jeffa> slide: NETLMM Problem Space
[11:15:25] <jeffa> describes the diagram
[11:15:32] <jeffa> slide: NETLMM Goals
[11:16:04] <dthaler> back looking at this jabber room now, are these slides online?
[11:16:13] <thomas.heide.clausen@gmail.com> Yes, they are.
[11:16:18] <jeffa> slide: NETLMM Diagram
[11:16:21] <thomas.heide.clausen@gmail.com> Autoconfiguration using DHCP slides: http://www3.ietf.org/proceedings/06jul/slides/autoconf-2.ppt -- presenter: Fred Templin
[11:16:32] <dthaler> thx
[11:16:38] <jeffa> oops : NETLMM Domain
[11:16:44] <jeffa> slide: NETLMM Using DHCP
[11:16:58] <thomas.heide.clausen@gmail.com> Unless I am really mistaken, this ppt would contain the two parts that Fred Templin presents.
[11:16:58] <jeffa> slide: Model of Operation
[11:17:43] <jeffa> slide: Model of Operation (cont'd)
[11:18:03] <jeffa> Fred is going to try and skip through as much of this as possible ... that is duplicated from the previous presentation
[11:18:18] <jeffa> slide: Route/Tunnel Conf... via AR1
[11:18:39] <jeffa> slide: Route/Tunnel ... moves to AR2
[11:18:57] <jeffa> slide: Additional Considerations
[11:19:04] <dthaler> Since I can't heard the voiceover, why are we talking about netlmm (I see slide 13 starts the "MANET Autoconfiguration using DHCP" presentation)
[11:20:01] <jeffa> first Fred presented about the MANET side of things, now he is talking about "what happens in the backhaul network"
[11:20:16] <thomas.heide.clausen@gmail.com> Dave:
[11:20:23] <jeffa> slide: Additional Considerations (cont'd)
[11:20:49] <thomas.heide.clausen@gmail.com> Basically, FT started with the MANET stuff, and then the transition was "Well, this somewhat fits with netlmm, so it might be a good idea to be aware what is going on there"....
[11:21:02] <jeffa> comments -- please put on the mailing list since we are running out of time
[11:21:15] <jeffa> THC: I don't have a lot to say, actually.
[11:21:31] <jeffa> what is important: firm promise to get AUTOCONF arch and PS docs done
[11:21:34] <jeffa> Joe Macker:
[11:22:03] <jeffa> JM: I think the problem that might make that more difficult is to coordinate what is within each ... time to make sure that is clean ...
[11:22:24] <jeffa> THC: we have had this week and I think there are more discussions scheduled, so this is important
[11:22:33] <jeffa> Jari: commenting on the coordination
[11:23:11] <jeffa> it is fine to coordinate, but I see a pattern here, the last time we had a discusssion here is April, it is fine to coordinate and discuss, everyone please work on the list, we need to get the two main documents out in order to continue
[11:23:27] <jeffa> THC: when we publish a -00 document it does not mean we publish something that is done
[11:23:36] <jeffa> it means it is something we are comfortable with
[11:23:53] <jeffa> we should get good coordination but not let that postpone the publication process
[11:24:09] <jeffa> we need to work together to get these docs into shape
[11:24:13] <jeffa> we will do this on the list
[11:24:36] <jeffa> JM: sounds great. one thing for the PS authors, there were some rough areas that were brought up even w/o coordination
[11:25:02] <jeffa> THC: I'm hoping PS authors will look at intermediate ... and perhaps PS and arch could get together this week.
[11:25:15] <jeffa> we need to get this out now...
[11:25:29] <jeffa> there are a lot of us that have been doing some autoconf-related activities
[11:25:46] <jeffa> now is the time to start writing that up and putting that out on the list
[11:26:12] <jeffa> I would consider it a huge failure if the arch document doesn't consider whatever your favorite deployment idea is.
[11:26:23] <jeffa> reality check on the solution domain
[11:26:55] <jeffa> the important part: those of you who are interested, talk to Emmanuel and Ian.
[11:27:13] <jeffa> let's get some emails out on the list
[11:27:39] <jeffa> I am depressed when I wake up and check the AUTOCONF list and see nothing
[11:27:40] <jeffa> any closing remarks/comments?
[11:27:56] <jeffa> thanks to the scribes
[11:27:56] <dthaler> yes good work on jabber scribing
[11:28:06] --- yangwoo ko has left
[11:28:20] --- badamson has left
[11:28:21] --- jeffa has left
[11:28:24] --- kazuya has left
[11:28:56] --- rjaksa has left
[11:28:59] --- monden has left
[11:29:14] --- thomas.heide.clausen@gmail.com has left
[11:30:32] --- yowada has left
[11:30:44] --- narten has left
[11:35:16] --- idc has left
[11:41:46] --- irino has left
[11:46:45] --- dthaler has left
[12:15:44] --- dthaler has joined
[12:15:48] --- dthaler has left
[12:37:52] --- irino has joined
[13:00:56] --- irino has left: Replaced by new connection
[14:05:35] --- thomas.heide.clausen@gmail.com has joined
[14:39:09] --- thomas.heide.clausen@gmail.com has left: Logged out
[15:17:50] --- jariarkko has left
[15:34:28] --- danjared has joined
[15:35:24] --- danjared has left
[17:15:07] --- LOGGING STARTED
[17:18:52] --- julien.garnier@gmail.com has joined
[17:52:07] --- julien.garnier@gmail.com has left: Logged out