[00:46:26] nnnshin joins the room [00:46:29] nnnshin leaves the room [02:44:18] mawatari joins the room [03:53:46] Shane Amante joins the room [03:58:21] Fuad Abinader joins the room [03:59:00] Simon Perreault joins the room [04:00:05] Bernie joins the room [04:00:33] Colman Ho joins the room [04:01:03] tetsuya.innami joins the room [04:01:41] Dave Thaler joins the room [04:01:56] becarpenter joins the room [04:02:00] mawatari leaves the room [04:02:08] nm joins the room [04:02:10] karen.s.seo joins the room [04:02:22] Agenda bashing started [04:02:26] arifumi joins the room [04:03:19] shamus joins the room [04:03:44] kivinen joins the room [04:03:49] kawashimam joins the room [04:03:53] nnnshin joins the room [04:04:43] described BoF scope [04:05:43] focus on Dual-Stack Lite encapsulation NOT select solution [04:05:56] define requirements [04:06:12] Ruri Hiromi joins the room [04:06:53] BoF intended to decide: Should IETF take on this work? Is the problem statement clear/solvable? Who is willing to help with charter text and write/edit documents? [04:07:00] Remis Despres: [04:07:15] discussing why restrict to DS Lite [04:07:25] tachibana@jabber.org joins the room [04:07:42] Christian Jacquenet: for expedience [04:07:52] Remi -- so where will the others be discussed? [04:08:09] Wing -- Maybe we can discuss further during writing of charter text [04:08:15] Margaret Wasserman [04:09:17] would like to not unduly restrict the tool(s) we develop [04:09:37] next talk -- A+P Overview (Pierre Levis) [04:10:06] mlepinski joins the room [04:10:08] Slide 2 Introduction [04:10:31] rdroms joins the room [04:10:48] will explain main ideas not to cover all proposals and use cases. [04:10:59] Slide 3 -- A+P names & proposals [04:10:59] Thanks for scribing from the room. [04:11:11] lots of different names, we'll use A+P [04:11:40] slide 3 (more text now showing) [04:11:56] lists a number of ways of implementing A+P [04:12:31] slide 5 -- Today Fixed Broadband Access [04:12:43] toshio.hiraga joins the room [04:12:43] satoru.matsushima joins the room [04:13:06] seems to be another version of slide 5 -- shows IPv6 on the line connecting the CPE NAT to the Internet [04:13:31] slide 6 -- A+P Port Range restriction [04:13:48] slide 7 -- shows restriction of port range [04:14:41] slide 8 -- shows "Here, PC's application are NOT Port Range aware!!!" [04:15:45] sorry I think the above slide was actually slide 10 [04:15:55] Slide 11 -- Outgoing packets [04:16:49] Lars Eggert: is there a NAT in the host? [04:16:56] Slide 13 -- "Port Range Handling of some sort" [04:17:01] Pierre: Either a NAT or all applications are changed [04:17:12] i didn't get why all applications would need to be changed [04:17:13] abaire joins the room [04:17:22] is this not a stack issue? [04:17:32] Slide 15 -- Solutions [04:17:59] Slide 17 -- PRR [04:18:11] slide 18 -- PRR [04:18:13] applications cannot use their favorite port number anymore [04:18:38] Slide 19 - PRR [04:18:47] CHENGANG joins the room [04:18:56] (uploaded slides are slightly different from the ones Pierre is using) [04:19:10] slide 20 with PRR [04:20:07] slide 20-23 -- PRR [04:20:34] arifumi: yes ok, only for not ephemeral ports [04:20:34] Magnus Westerlund [04:20:53] speaker seems to have slide 24 up [04:21:09] shep joins the room [04:21:11] and slide 25 [04:21:50] seems very hot in here. If it is not just me, is someone being notified so something can be done? [04:21:57] margaret wasserman [04:22:15] slide 26 -- Example2: A+P with an IPv4-IPv6 Gateway for Fixed Broadband Access [04:22:17] even one of the japanese ppl agrees its hot here. :) [04:22:42] slide 28 [04:22:55] slide 29 [04:23:08] I'll see what I can do. [04:23:29] slide 30 [04:23:37] slide 31 [04:23:55] Colin Perkins [04:24:24] IETF staff and building maintenance has been notified about the temp in the room. [04:24:33] Thank you [04:25:07] Thank you. Feels like it is getting a little cooler already. (Maybe it is just in my mind.) [04:25:09] don't know who this is [04:25:10] ooooh, fresh air! [04:25:21] goood [04:25:45] tonyhansen joins the room [04:26:28] Speaker agrees there are scenarios where there are NATS instead of IPv4/IPv6 gateways. [04:26:33] Margaret Wasserman [04:27:19] The host, and the CPE, are IPv4-only in this scenario. [04:27:26] Remi Despres [04:27:43] the CPE is not IPv4-only in this cenario, it has to do 4 in 6 encap [04:28:09] Oh, in the current slide, yes, but I was further back [04:28:51] one of the benefits of the A+P is the resulting public address [04:28:55] Margaret Wasserman [04:29:37] Chris Donley joins the room [04:29:52] Remi Despres [04:30:55] Margaret Wasserman [04:31:18] Yao Li (??) [04:32:00] Colin Perkins [04:32:49] Lars Eggert [04:33:35] Wing - [04:33:39] Remi [04:35:24] wants both IPv4 and IPv6 connectivity for his hosts [04:35:28] hirocomb joins the room [04:35:38] Who is the speaker? [04:35:47] Remi [04:35:48] Lars and Remi are speaking [04:36:02] Li XIng [04:36:32] Could replace the CPE NAT with port mapping [04:36:55] gigix73 joins the room [04:37:04] i guess its also called nat usually :) [04:37:50] slide 33 [04:38:24] colin [04:38:46] speaker [04:39:11] colin [04:39:24] stuart cheshire [04:40:24] alain durand [04:41:24] speaker [04:41:34] Bernie leaves the room [04:41:44] arifumo matsumoto [04:42:13] Bernie joins the room [04:42:32] HUI joins the room [04:42:36] wing: assuming new version [04:42:41] remis despres [04:43:59] shin miyakawa [04:44:08] hirocomb leaves the room [04:44:50] slide 9 in uploaded version [04:45:03] wing -- this is out of scope for this BoF [04:45:29] lars eggert [04:45:58] Ed J. joins the room [04:46:09] remi despres [04:46:19] HUI leaves the room [04:46:50] speaker ?? [04:47:09] colin perkins [04:48:08] Sri Gundavelli [04:48:29] Wing -- wait for mobile presentation [04:48:35] Margaret Wasserman [04:49:31] remi despres [04:50:40] margaret [04:51:11] remi [04:52:04] I don't understand the 'stateless' claim in the case of an inbound packet that has been fragmented. The ISP box will need to reassemble it before forwarding it to the appropriate customer (and maybe fragment it again in the worst case). [04:52:22] right [04:52:33] shinmiyakawa joins the room [04:52:47] next speaker -- Behcet Sarikaya [04:52:59] I suspect he's thinking about per-flow state rather than per-packet state [04:53:05] +1 to Brian. We'll have more expert insight from DaveT shortly. [04:53:14] A+P for dual-stack mobile IPv6/Proxy Mobile IPv6 [04:53:34] per-flow state in the "usual" case; not considering corner cases. [04:53:40] I won't talk about stateless per se (since I don't see it any different from double NAT, and I'm talking about things that are different) [04:53:45] @Dave: yes, there's probably no per flow state [04:53:46] slide 2 -- context and motivations [04:54:31] slide 3-- Context and motivations [04:54:48] I've heard that IPv4 fragmentation will become more common with DNSSEC, and we will soon care that it is working right (including reassembly) everywhere. [04:55:34] slide 4 -- Context and motivations [04:56:12] yes shep [04:56:22] ASHIDA joins the room [04:56:50] slide 5 -- Context and motivations [04:58:02] slide 6 -- Conclusions [04:58:06] Alain Durand [04:58:44] Dave Thaler [04:58:47] gigix73 leaves the room [04:58:51] Desire joins the room [04:59:24] JariArkko [04:59:56] @Dave Thaler Then, my question at previous presentation is IN the scope isn't it ? :-) [05:00:20] @shin: in my opinion, yes [05:00:24] speaker is unknown to me [05:00:33] Sri Gundavelli [05:00:52] evyncke joins the room [05:00:58] Alain Durand [05:01:40] [05:01:59] missed identity of last speaker [05:02:04] Greg Lebowitz [05:02:16] Greg is now [05:03:10] Remi Despres [05:04:41] Sri Gundavelli [05:04:48] Remi [05:05:18] Yao Li [05:06:16] next speaker -- Dave Thaler [05:06:22] "Issues with Port-Restricted IPs" [05:06:37] Slide 2 -- Scope [05:07:44] Slide 3 -- General Issue [05:09:17] Alain Durand [05:09:44] Thaler [05:09:48] touch joins the room [05:09:58] Slide 4 -- Implementation: hosts (1/2) [05:11:51] Bruno STEVANT joins the room [05:12:57] well,,,, a+p is here to avoid NAT, isn't it ? [05:13:36] Slide 5 -- Implementation: hosts (2/2) [05:13:47] it's to avoid the CGN, not the NAT on customer premises [05:14:05] looks like NAT is un-avoidable [05:15:15] slide 6 -- Management: ping [05:16:06] I assumed that NAT on customer premises are transitional workaround before host's A+P support. [05:16:14] slide 7 -- Other non-port based protocols [05:17:30] HUI joins the room [05:17:38] slide 8 -- Provisioning system [05:17:58] behcet.sarikaya joins the room [05:18:03] touch leaves the room [05:18:54] slide 9 -- Training/education [05:20:07] slide 10 -- Security [05:21:20] slide 11 -- Failure modes [05:21:26] CHENGANG leaves the room [05:22:33] slide 12 -- Long-term impact [05:24:06] slide 13 -- Summary [05:24:53] tachibana@jabber.org leaves the room [05:25:56] 30 minutes of open discussion -- Wing [05:26:11] is dave's slide avilable? [05:26:13] @Dave Thanks ! Tereffic. You've answered all I have . [05:26:59] masataka ohta [05:27:58] http://www.ietf.org/proceedings/09nov/slides/aplusp-3.pdf [05:28:39] Masataka Ohta talking about his own host-NAT proposal [05:28:55] what draft? [05:29:01] Alain Durand [05:29:11] end-to-end nat or sometihng like that [05:29:29] hirocomb joins the room [05:29:37] thanks [05:29:40] draft-ohta-e2e-nat-00 <-- this? [05:29:48] where is randy ? [05:30:13] i guess it is. [05:30:20] Alain -- the history is that Randy bush invented to maintain control with the end users; maybe there are limited cases where this can be used [05:30:31] Yes, draft-ohta-e2e-nat [05:31:01] Alain -- should not be a new WG, better to take some of this work in an existing WG to solve specific problems [05:31:19] Remi Despres -- agree with Alain and SOFTWIRE would be a good place [05:31:39] Remi -- also disagrees with Dave Thaler and would happily discuss it offline [05:31:51] Margaret Wasserman [05:33:36] I don't understand what the working would do [05:33:43] (the last was from Margaret) [05:33:54] Ralph Droms [05:35:36] Ralph: Why a WG? We have a proposal for an additional technology that will help with IPv4/IPv6 co-existence and with IPv6 deployment. There are outstanding problems that would need to be addressed. The WG would examine the application of A+P to Dual-stack Lite. [05:36:09] Bernie leaves the room [05:36:20] Alain Durand [05:36:36] A+P wouldn't replace the CGN after all? count me among the confused. [05:37:00] Alain -- one place in Softwire Dual-stack Lite and in appendix where this might be used. Better to not create another working group [05:37:28] Ralph Droms - But just because work starts in a given WG doesn't mean it should stay there. [05:37:58] so it does replace the CGN after all [05:38:06] not confused anymore, i think [05:38:34] Ralph: This proposal moves this aspect of the NAT from the service provider core to CPE. Whether or not it actually does this is one of the topics of discussion. [05:38:48] Lorenzo Colitti [05:39:17] Lorenzo -- what do net operators think? what do they think about 6rd? [05:39:38] Pierre Levis [05:40:31] Bernie joins the room [05:40:41] Pierre -- I guess a lot of points in Dave's talk are debatable. But I'd like to come back to the case with the CPE -- putting a router in with a new feature. I've talked to all the router vendors and doing this is not hard. [05:40:53] (speaker?) [05:41:07] Julien laganier [05:41:37] abaire leaves the room [05:41:38] Lars Eggert [05:41:38] anthony.baire joins the room [05:41:43] Remi Despres [05:43:47] tonyhansen leaves the room [05:43:52] Remi -- the origination of A+P is in the work of people from different countries, major telecom, vendors, etc. [05:44:00] Masataka Ohta [05:44:39] Masataka: A+P working group is good idea but should be neutral re: IPv6 [05:45:03] Bob Hinden -- numerous people ask me which of many transition mechanisms should they use. [05:45:04] shep leaves the room [05:45:10] PasiS joins the room [05:45:42] Bob -- need less complexity [05:45:52] Colin Perkins [05:46:41] Colin: A number of people said this is easy to implement. But as Dave's presentation described, it has massive implications. I agree with Lars Eggert --> really bad idea [05:46:53] Xavier Marjou [05:47:30] Xavier -- We should document these things including the limitations. [05:47:37] Jari Arkko [05:48:30] Jari -- agree with Bob -- should limit complexity. But... if you're asking about a new WG -- I'd say no. If you're asking about an experimental change in context of Dual-Stack lite, that may be fine. [05:48:45] Philip Eardley [05:49:29] Satoru Matsushima [05:49:46] Ed J. leaves the room [05:50:08] Masataka Ohta [05:50:55] Masataka -- make it clear that x maintains end-to-end transparency [05:51:12] x = his proposal again [05:51:27] Greg Lebovitz -- Better to take the energy that would be spent on this and spend it on patching hosts [05:51:42] Markus Isomaki [05:52:24] Markus: this may have value vs double-NAT'ing. Keeping the narrow scope, this probably makes sense. [05:52:30] Remi Despres [05:53:15] Remi -- presenting this in Softwire as part of 6rd would be adequate [05:53:20] it's a very very small wording change to the definition of IP... [05:53:34] which has a huge impact [05:53:52] Chris Donley leaves the room [05:53:53] Lars Eggert -- horribly complicated, even in just Dual-stack Lite. We don't need it. [05:54:08] Remi Despres -- let those who want to implement it [05:54:11] donley.chris joins the room [05:54:21] Masataka Ohta [05:54:47] Masataka -- have implementation working except for ping [05:55:04] Ralph Droms [05:55:37] This is an opinion gathering situation for input to IESG, not a consensus vote [05:56:09] 1. Is the problem statement clear? [05:56:15] louder for not clear [05:56:24] Is the problem statement solvable? [05:56:35] louder for not solvable [05:56:42] tetsuya.innami leaves the room [05:57:16] 3. should the work be taken on? (how is separate question) [05:57:38] behcet.sarikaya leaves the room [05:57:46] What are you humming on if you didn't understand the problem statement? [05:58:48] Who thinks work involving extensions to DS-LIte for A+P should be taken on? [05:59:00] Greg Lebovitz speaking [05:59:08] Ralph Droms [05:59:11] evyncke leaves the room [05:59:26] nnnshin leaves the room [05:59:46] Ralph: The scope of the BoF is A+P in the context of DS-Lite. [05:59:52] shep joins the room [06:00:45] hum for no was louder [06:00:51] Ruri Hiromi leaves the room [06:01:01] becarpenter leaves the room [06:01:05] kawashimam leaves the room [06:01:07] Bruno STEVANT leaves the room [06:01:08] Simon Perreault leaves the room [06:01:09] Lars -- Is this going to be discussed in Softwire? [06:01:11] toshio.hiraga leaves the room [06:01:13] mlepinski leaves the room [06:01:14] PasiS leaves the room [06:01:17] shamus leaves the room [06:01:31] (That is up to the IESG to determine) [06:01:35] satoru.matsushima leaves the room [06:01:37] shinmiyakawa leaves the room [06:01:45] Bernie leaves the room [06:01:58] ASHIDA leaves the room [06:02:02] arifumi leaves the room [06:02:02] kivinen leaves the room [06:02:04] Shane Amante leaves the room [06:02:05] donley.chris leaves the room [06:02:07] rdroms leaves the room [06:02:33] Fuad Abinader leaves the room [06:02:58] Desire leaves the room [06:04:54] HUI leaves the room [06:05:37] hirocomb leaves the room [06:08:06] Colman Ho leaves the room [06:08:59] karen.s.seo leaves the room [06:09:30] anthony.baire leaves the room [06:13:52] nm leaves the room: Replaced by new connection [06:14:29] Ruri Hiromi joins the room [06:14:50] anthony.baire joins the room [06:14:59] shep leaves the room [06:15:31] Dave Thaler leaves the room: Replaced by new connection [06:30:43] Bruno STEVANT joins the room [06:30:49] Bruno STEVANT leaves the room [06:31:19] anthony.baire leaves the room [06:31:20] anthony.baire joins the room [06:34:45] Colman Ho joins the room [06:38:44] PasiS joins the room [06:40:28] PasiS leaves the room [06:44:28] Colman Ho leaves the room [07:11:17] Ruri Hiromi leaves the room [07:25:21] Colman Ho joins the room [07:25:35] Colman Ho leaves the room [07:32:09] anthony.baire leaves the room [07:33:41] Ruri Hiromi joins the room [07:54:10] arifumi joins the room [07:54:12] arifumi leaves the room [08:28:23] Bruno STEVANT joins the room [08:28:33] Bruno STEVANT leaves the room [10:49:39] Ruri Hiromi leaves the room [15:12:21] Ruri Hiromi joins the room [22:43:33] Ruri Hiromi leaves the room