[20:44:12] Seung joins the room [21:25:01] cjbernardos joins the room [21:29:51] meeting starts [21:29:58] jariarkko joins the room [21:30:05] Charlie takes notes [21:30:15] Dave Thaler and myself javver scribes [21:30:19] narten joins the room [21:30:46] Has the session started? I get no audio yet... [21:30:51] yes [21:31:00] use the mikes please... [21:31:00] it's about to start [21:31:19] agenda [21:31:26] http://www3.ietf.org/proceedings/08nov/agenda/autoconf.txt [21:31:34] I can't hear... [21:31:48] he is speakin to the mikes. [21:31:54] loudspeakers are on in the room [21:32:02] they are not working well ... cannot hear [21:32:09] I barely hear someone talking... [21:32:12] jari is taking care of this [21:32:50] shubranshu is summarising the status [21:32:58] jari testing the mikes [21:33:06] I hear jari, fine, but not the chairs [21:33:08] I spoke to the other mic [21:33:11] thclausen joins the room [21:33:21] Narten, do you hear ? [21:33:27] he is speaking more loudly now [21:33:36] can barely hear, but now can make it out. [21:33:43] it is still way lower than the other mike [21:33:53] yes! much better! [21:33:59] Shubhrnshu changed to speaker mike. [21:34:04] :-) [21:34:11] So the chair mike is not working - good stuff. [21:34:20] DThaler joins the room [21:34:25] thomas presentation [21:34:47] (no slides on the IETF meeting materials site) [21:35:04] there's 31 people physically in the room [21:36:07] yep, no slides for this talk.. :-( [21:36:07] shubranshu joins the room [21:37:53] historical and background review of the draft [21:38:02] status of the draft: individual submission [21:38:35] content: [21:39:11] background stuff: manet "node taxonomy" (coming from 67th IETF) [21:40:24] separation of manet node into a router component and a host component isolated from manet interface [21:41:05] manet interfaces -slide [21:41:21] remote folk do not have the slides... they were not updloaded... [21:42:05] thomas is uploading the slides now [21:42:11] Sorry, folks. [21:42:30] nordmark joins the room [21:42:55] slides coming to shubhransu, he will upload them [21:43:24] slide shows a topology with a series of 7 manet nodes in a line [21:43:42] each one can only reach the adjacent ones, and the two ends wants to communicate [21:44:34] http://www3.ietf.org/proceedings/08nov/slides/autoconf-2.pdf [21:44:35] got the presentation. thanks! [21:46:05] network dynamics -slide [21:46:05] network dynamics slide [21:46:22] lion joins the room [21:46:22] now on 21/38 [21:46:44] (since multi[ple have the same title :) [21:46:57] 22 [21:47:04] thanks... it is hard to figure out which sub page to be on. [21:47:19] Shaun Kaplan joins the room [21:47:31] 23 [21:48:27] 24 [21:48:39] this guy = N3 [21:49:42] charlie on the mike [21:50:11] dave thaler on the mike [21:52:30] thaler: the iab document defines link as something where ttl need not be decremented. clausen's use is more like what people normally call "segment", i.e., l2 technology related notion [21:53:28] erik asks about the desired behaviour. what do we want to do, and what do applications assume? [21:54:22] 25 [21:54:54] 26 [21:55:15] 27 [21:55:35] 29 [21:55:53] 30 [21:56:03] 31 [21:56:18] 32 [21:57:47] we are still being loose with terminology, and in particular, whether we are talking aboutIP-level semantics, or data-link (L2) level properties. [21:57:56] This is Seung Yi of Boeing. Could you clarify the last bullet on the slide 32? [21:58:10] nordmark leaves the room: Replaced by new connection [21:58:11] nordmark joins the room [21:58:40] seems to me, neighborcast is an L2 statement, not an L3 statement. IP has no notion of neighborcast. [21:59:12] I have a problem with saying IP LL multicast/broadcast becomes "neighborcast" [22:00:00] Thanks, Dave. [22:00:58] I think he means LL multicast might not even be able to get to everyone withing TTL1? (not sure) [22:01:15] neighborcast is an L2 property. To say it is visible to IP level is saying you are changing the IP model. [22:02:25] Jari is on to my point. I think we are really talking about a layer 2.5. something below IP, but above l2. [22:03:07] Wouldn't it be better to introduce a new term to replace the "Link" and keep the definition of the older "link" intact? Perhaps, it's time to consider bringing back "site"? [22:04:12] how about "unknown cast". :-) [22:04:17] :) [22:04:49] "neighborcast" is something larger than "link" (== TTL 1) and smaller than "site" or "MANET Link" [22:04:57] OSPF is funny, it is one application. [22:05:02] there's other scop numbers in IPv6 multicast :) [22:06:38] you need to distinguish between how OSPF works over an NBMA interface and how IP works over such an interface. [22:06:45] 36 [22:07:38] for things like ND and ARP and linklocal b/mcast, NBMA links emulate that functionality. [22:07:39] (actually I take back what i said above, I'm not sure if "neighborcast" is larger or smaller or equal to TTL=1) [22:07:55] well SOME nbma links emulate that functionality :) [22:08:02] which ones don't? [22:08:18] depending on how loose/strict you are with "emulate" [22:08:28] some do algorithmic mapping with no messages (6to4 for instance) [22:08:50] they have a way of providing the key functions (address resolution, etc.) [22:09:27] yes, that's the loose interpretation, sure I agree [22:10:36] how is speaking at mike (don't recognize voice) [22:10:41] but in a way, the adhoc guys do provide a level of emulation -- by separating the special link from the ones that have actual hosts. [22:10:53] Zach something? [22:11:23] jari: correct. so if you view a MANET is only having routers on it, all you have to do is figure out how to provide address resolution and a few other properties. You may not need to make multicast work. [22:11:55] Zach Shelby was his name [22:12:09] kind of like NMBA. only need to make it work enough so routers can use it... normal hosts doing multicast and other standard things won't be using the link directly. [22:14:20] dow street on the mike [22:15:58] vifajardo1 joins the room [22:17:22] (slide 36) [22:17:55] lion leaves the room: Computer went to sleep [22:18:28] lion joins the room [22:20:44] he is referring to the "R" routers in slide 36 [22:21:31] LIfe would be a lot simpler if we just said: all links are point-to-point (with neighbors in radio range), and you then do flat routing to handle movement. [22:21:58] I gather that is one possible model, but not the only one... [22:22:03] agree [22:22:38] we presumably know how to make flat routing with point-to-point links work. [22:26:13] I believe we do Thomas N. That's exactly how we built a MANET autoconf prototype in Boeing and it works fine. Only remaining hole we had was the host mobility between MANET routers, which we circumvented by relying on HIP. [22:27:10] slide 14 [22:27:17] well, this WG doesn't seem to be able to agree on anything, even one simple scenario. We always seem to want to leave things flexible, but that introduces complexity and it is unclear (to me) what the benefit is. I agree strongly with what Erik is saying, btw. [22:28:33] It would be a big step forward to pick the simple model as one model, and defining it. Doing so, would not preclude other models. But by never "dividing and conquering" the over model/solution space, we remain wedged. [22:29:33] I think the discussion is again between "link" and "Link". Two different concepts. [22:30:08] ("IP link" vs "MANET Link" :) [22:30:21] (slide 27) [22:30:33] intvelt joins the room [22:30:37] we have to be precise about terminology. Otherwise we will continue to talk past each other (like we seem to be doing now...) [22:32:28] Shaun Kaplan leaves the room [22:32:29] thomas narten, your comments are on their way to the mike. [22:32:54] Is the queue still long (and increasing?) :-) [22:34:51] the queue has been confused. its in better alignment now, I think [22:41:15] (slide 38 :-) ) [22:46:36] people attending remotely: let us know if you need us trying to summarise the discussion (in case the audio streaming is not clear) [22:47:00] Audio streaming works perfect for me. Thanks. [22:47:12] (slide 26) [22:48:30] audio streaming has been superb so far this week.... [22:49:26] (great, it'd been hard to summarise all of what is being discussed :-) ) [22:51:22] can't hear the question [22:52:15] Thomas may not think so, but I think this document has been helpful in getting us to the point where we seem to be in agreement today. It was not a waste of time at all! [22:52:33] shubranshu is asking if there is anybody against documenting the manet link model [22:52:56] There are many IETF documents where the document itself is not that useful when all is done, but the process that led to the document was critical. [22:54:38] Someone remind THomasC that this document is means to an end, it is not the end itself. [22:57:00] (I think Thomas C is fully aware of that) [22:57:24] Keep in mind that one of the reasons we have been having this discussion about the manet "link model" is becuase folk were starting out by saying "ND doesn't work over Manet". But ND does work just fine over the point-to-point model we seem to be accepting. [22:58:13] charlie is arguing AGAINST a link being a white circle :) [23:03:57] we need to not interchange "link" and "interface". [23:04:06] they are different terms, IMO. [23:04:28] +1 narten [23:07:50] meeting finishes [23:07:50] nordmark leaves the room [23:08:09] lion leaves the room [23:08:18] (apologies if I couldn't get to the mike in time for all the comments) [23:08:45] nordmark joins the room [23:08:55] Thanks for scribing. :) [23:09:10] Seung leaves the room [23:09:16] thclausen leaves the room: Logged out [23:11:06] cjbernardos leaves the room [23:16:32] intvelt leaves the room [23:18:18] shubranshu leaves the room [23:25:25] DThaler leaves the room [23:30:29] nordmark leaves the room [23:31:05] nordmark joins the room [23:33:58] jariarkko leaves the room [23:54:09] DThaler joins the room