[17:08:48] cbyrne joins the room [18:39:25] cbyrne leaves the room [19:20:04] cbyrne joins the room [19:21:18] cbyrne leaves the room [19:22:12] cbyrne joins the room [19:33:30] cbyrne leaves the room [19:36:49] behcet.sarikaya joins the room [19:42:40] cbyrne joins the room [19:48:56] WesG joins the room [20:01:34] Dan Wing joins the room [20:02:20] Dan Wing has set the subject to: BEHAVE, IETF77 [20:03:36] arifumi joins the room [20:03:43] Lars joins the room [20:03:53] JonathanLennox joins the room [20:04:14] tskj joins the room [20:04:23] محسن joins the room [20:04:53] skudou joins the room [20:05:07] billvs joins the room [20:05:24] jpc joins the room [20:05:48] 2758746fb9640fab joins the room [20:05:49] ikuhei joins the room [20:06:08] magnus joins the room [20:07:12] Hiromi Morita joins the room [20:07:59] Scott Brim joins the room [20:08:18] Andrew joins the room [20:08:24] cbyrne leaves the room [20:08:26] ha, at last [20:08:31] I'll be a jabber scribe [20:08:35] sorry it took so long [20:09:05] Now on Magnus Westerlund's slides about 5 translator docs [20:09:08] slide 3 [20:10:00] cbyrne joins the room [20:10:12] Shane Amante joins the room [20:10:21] cbyrne leaves the room [20:11:14] looking for input . . . [20:11:23] objection to MUST in this context? [20:11:46] [couldn't hear] at mic: stateful case? [20:12:23] Clarification on MUST: [20:12:35] shinmiyakawa joins the room [20:12:46] cbyrne joins the room [20:12:49] Back to Magnus [20:13:11] [no name] can't work with stateful because it needs ports [20:13:28] Note: I can't look up and type at the same time, because I'm incompetent [20:13:41] so if you don't say your name clearly at the mic, it won't get in here [20:13:52] Marcelo Bagnulo @ mic [20:13:58] [our slightly tweaked agenda is now at http://www.ietf.org/proceedings/10mar/agenda/behave.html The change was to swap Dean Cheng and Xiaohu Xu's presentations, so Dean's is now first.] [20:14:29] tsavo_work@jabber.org/Meebo joins the room [20:15:36] Scott Brim leaves the room [20:15:56] Marcelo @ mic again [20:16:14] Suz joins the room [20:16:43] [unnamed] what do 44 NATs do? [20:16:49] Next Steps slide [20:16:51] Simon Perrault [20:17:07] is answering [20:17:19] Atarashi Yoshifumi joins the room [20:17:33] satoru.matsushima joins the room [20:17:52] Phil argued for SHOULD [20:17:57] Simon argued can't be MUST [20:18:39] [unnamed] Box isn't only NAT [20:19:00] Marcelo again: going to send ICMP error back if drop, right? answer: yes [20:19:01] unnamed was Toerless Eckert [20:19:18] slide 4 [20:20:27] slide 5 [20:21:20] [unnamed sounds like Fred Baker] question @ mic: does this require state? [20:21:31] yes. [unnamed] No. Plain no. [20:21:49] yes that was Fred Baker :) [20:21:54] just saying no. [20:22:16] tinnami joins the room [20:22:32] slide 6 [20:22:53] Colman Ho joins the room [20:23:01] slide 7 [20:23:01] Scott Brim joins the room [20:23:42] Dave Thaler responding to Fred's earlier question [20:24:39] More comments? [20:24:46] slide 8 [20:24:58] Paul Selkirk joins the room [20:26:12] Dan Wing setting stage [20:26:21] Purpose of Today slide [20:26:45] http://www.ietf.org/proceedings/10mar/slides/behave-13.pdf slide 10 [20:27:36] Please check the wiki [20:28:24] Dave Thaler: if you think there's work, that doesn't mean it needs to be standardized [20:28:47] Wiki is at http://trac.tools.ietf.org/wg/behave/trac/wiki/WikiStart [20:28:49] sean.s.shen joins the room [20:28:58] next slides [20:29:16] http://www.ietf.org/proceedings/10mar/slides/behave-1.pdf [20:29:20] slide 2 [20:29:43] slide 3 [20:30:40] becarpenter joins the room [20:30:52] wmhaddad joins the room [20:31:38] slide 4 [20:31:39] slide 5 [20:32:49] slide 6 [20:34:27] slide 7 [20:35:35] Marcelo Bagnulo at mic [20:36:15] Phillip Matthews [20:37:53] Alain Durand [20:40:21] Ashida joins the room [20:40:22] Remy [20:40:26] then Simon Perrault [20:40:28] then I didn't hear [20:40:36] (sorry, all, interrupted) [20:40:41] sean.s.shen leaves the room [20:40:49] Back to slide 7 [20:41:10] slide 8 [20:41:47] slide 9 [20:41:51] HUI joins the room [20:42:30] Ashida leaves the room [20:42:44] 芦田 joins the room [20:43:48] [Lars Eggert?] at mic [20:44:14] Scott Brim leaves the room [20:45:08] Dan asks for a summary of the problem that this is solving [20:45:28] Back on slide 3 [20:46:32] mhasib joins the room [20:47:09] Long line at mic [20:47:59] Dan Wing suggesting words for speaker [20:48:08] [unnamed] @ mic [20:48:14] back to slide 9 [20:48:48] It was Gregory Lebovitz [20:49:57] mhasib leaves the room [20:50:06] Marcelo @ mic [20:50:55] Philip Matthews [20:51:27] showing slide 4 [20:52:01] Alain Durand [20:52:49] skipping over slides 10 and 11 [20:52:53] slide 12 now [20:53:18] slide 13 [20:54:32] Chairs cutting off remainder -- out of time [20:55:05] Simon Perrault @ mic [20:55:21] tinnami leaves the room [20:56:12] Gregory @ mic again [20:56:23] Alain @ mic again [20:57:04] Simon again [20:57:28] Next preso [20:57:45] http://www.ietf.org/proceedings/10mar/slides/behave-3.pdf [20:58:00] Gregory joins the room [20:58:02] slide 3 [20:58:08] slide 4 [20:58:27] if i understood correctly, [20:59:03] the last solution provides benefit because, in the best case where hosts use the range of ports allowed by the NAT for it, it is completely stateless [20:59:06] slide 5 [20:59:33] slide 6 [20:59:38] and in the worst case, [21:00:00] when it uses a port outside of the range allowed it by the NAT, then the NAT must keep state [21:00:18] it simply reduces the total number of sessions that must be statefully kept [21:00:31] here's another thought: [21:00:32] [ [21:00:37] a bit [21:00:41] yes [21:00:47] another thought: [21:01:02] if the port used is outside the allotted range, nat could drop packet [21:01:27] jinmei joins the room [21:01:35] Also it breaks 5452 because end nodes won't have enough ports to get a decent range from which to choose. [21:01:36] when hosts sees connection is failing, it tries another port, and does so until it finds one within its allocated range [21:01:40] slide 8 [21:01:51] slide 9 [21:02:21] slide 10, then 11 [21:02:32] and if u think that won't work because hosts can't do that well, then look at skype, and flash, which both do the same [21:02:49] in that case, we could get rid of the partial state scenario, and go fully stateless [21:02:58] thoughts? [21:04:29] I guess I can't tell how many ports hosts are going to get. The point of DNS port randomization is to get additional randomness so that DNS answers are hard to spoof. If you reduce the number of ports available to just a few, this answer is broken. [21:04:45] http://www.ietf.org/proceedings/10mar/slides/behave-9.pdf [21:04:49] Next presentation [21:04:57] slide 2 [21:05:21] 芦田 leaves the room [21:05:32] Ashida joins the room [21:05:35] [The Partial State proposal does not reduce the number of ports the host can use.] [21:06:39] @Andrew: remember this is a TRANSITION mechanism. It doesn't have to be an ideal end state. I just has to get us all across the chasm, then we can burn it (sure, 30 yrs from now... but we don't have to live with it forever) [21:06:42] tinnami joins the room [21:06:45] [The Partial State proposal is stateless if the host uses the NNN ports allocated to its partial state use; that is, it effectively does Stateless NAT64. If the host uses a port that is _not_ in that range for partial state, it gets Stateful NAPT64.] [21:07:10] right [21:07:11] got it [21:07:13] [this is best summarized, I believe, on Slide 9 of the Partial State presentation.] [21:07:13] Well, the immediate result will in practice be that people want to live inside their "native" port range, because all that state for DNS queries is expensive. If the number of such ports is small, then the randomness is reduced. [21:07:34] I'm trying to see if we can get rid of the case where stateful NAPT64 is required at all [21:07:39] slide 3 [21:07:49] slide 4 [21:07:51] jinmei leaves the room [21:07:55] i.e. if port P used is not in set of NNN range, then drop packet [21:08:00] Marc Blanchet joins the room [21:08:10] and host will retry on another port until P is in set of NNN [21:08:13] slide 5 [21:08:23] [Gregory: Maybe could do that. But that is not the Partial State proposal.] [21:08:26] a totally different take on it, in order to go completely stateless [21:08:33] correct [21:08:38] I'm being creative ;-) [21:08:41] slide 6 [21:08:54] Philip Matthews at mic [21:09:01] Gregory is wondering if creative is still in scope? [21:09:56] wmhaddad leaves the room [21:09:59] using slide 5 to answer [21:10:14] Philip again [21:12:13] Dave Thaler [21:12:20] Philip again [21:13:15] jinmei joins the room [21:14:23] Jinmei just now? [21:14:29] then Xing, I think [21:15:14] JonathanLennox leaves the room: Computer went to sleep [21:16:06] Andrew Sullivan [21:16:22] then Jinmei again [21:16:30] Alain Durand [21:16:31] no [21:16:37] Charles Perkins [21:17:36] Alain Durand [21:18:12] tinnami leaves the room [21:18:39] Macelo Bagnulo [21:19:31] JonathanLennox joins the room [21:19:55] HUI leaves the room [21:20:06] next NAT Synchronization using SCSP [21:20:50] slide 4 [21:20:55] (grr. No line numbers) [21:21:00] (err, page numbers) [21:21:03] velt joins the room [21:21:17] slide 6 [21:21:32] slide 7 [21:21:49] chris liljenstolpe joins the room [21:22:14] tsavo_work@jabber.org/Meebo leaves the room [21:22:43] jpc leaves the room [21:22:52] slide 8 [21:24:10] slide 9 [21:25:33] slide 10 [21:25:43] Dave Thaler no hat [21:27:52] Lars Eggert [21:28:16] donley.chris joins the room [21:28:28] lars is way outside his area of exposure here [21:28:59] [I didn't catch that name, sorry] [21:29:07] speaking now is Philip Matthews [21:29:13] Oh, wait, it's Philip [21:29:20] thanks [21:32:24] [missed last name[ & now Gregory Lebovitz [21:32:48] tsavo_work@jabber.org/Meebo joins the room [21:34:13] Hadriel Kaplan [21:35:26] Bob Hinden [21:37:14] http://www.ietf.org/proceedings/10mar/slides/behave-6.pdf [21:37:30] slide 2 [21:37:38] slide 3 [21:39:21] slide 4 [21:39:23] slide 5 [21:39:40] slide 6 [21:39:49] slide 7, 8, 9 [21:39:51] slide 10 [21:40:04] slide 11 [21:41:10] Hadriel Kaplan @ mic [21:42:33] Gregory Lebovitz @ mic [21:43:19] [missed name] [21:43:22] Shin [21:45:31] Ronald van der Pol joins the room [21:45:42] Shane Amante[?] [21:45:43] Colman Ho leaves the room [21:46:16] Colman Ho joins the room [21:47:30] I, Shin Miyakawa was the person who next to Gregory [21:48:03] http://www.ietf.org/proceedings/10mar/slides/behave-12.pdf [21:48:05] slide 2 [21:48:08] Thanks! [21:49:27] slide 4 [21:50:18] slide 5 [21:50:22] slide 6 [21:50:51] Brian Carpenter [21:51:46] Hadriel Kaplan [21:52:15] billvs leaves the room [21:53:15] IMHO -- slide 7, no need to optimize the host in light of the fact that slide 6 is often the case. [21:53:34] Sounds like Fred Baker at the mic [21:53:39] but he didn't say so [21:53:49] Atarashi Yoshifumi leaves the room [21:54:22] Sounds like maybe Alain Durand [21:54:43] @ Andrew: kudos for spelling my name right in the jabber scribe. thx!! ;-) [21:54:47] that's Francis Dupont [21:54:51] last name that is [21:55:21] Ah, yes, it was Francis, thanks Suz [21:55:27] Didn't catch who this is either [21:55:42] fdupont joins the room [21:55:59] @Gregory: it's not hard: I just go & fetch people's names from list postings :) [21:56:05] This is Alain Durand now [21:56:44] back to slide 2 to answer this [21:57:12] satoru.matsushima leaves the room [21:57:24] Explaining my idea: today the whole stuff relies on a central DNS64 server which magically (aka synthesis) redirect traffic to translate to the NAT64. [21:57:58] I couldn't hear any of that [21:58:02] so I didn't catch the name [21:58:04] Hadriel again [21:58:23] This is bad for at least two reasons: synthesis is broken with DNSSEC because it is a form of lie. The dual stack issue exposed is another. [21:58:25] btw, [21:58:38] did anyone get the NAT64 network working for them here at ietf? [21:58:54] ie. by connecting to the ietf-nat64 ssid? [21:59:08] I tried during the (thrilling) plenary last night, [21:59:18] when I wasn't rivetted by the day-pass debate, [21:59:23] My proposal is to make hosts more aware so the DNS64 server becomes a fallback mechanism (for unaware/dumb hosts) and not a lying/misleading one. [21:59:26] and couldn't get it going on my mac [21:59:36] did it work for other behave'rs? [21:59:38] I didn't get the last mic person [21:59:46] Now Erik Nordmark [21:59:59] I got it working [22:00:03] [22:00:09] Now Alain Durand [22:00:16] [22:00:19] you need two things: [22:00:23] billvs joins the room [22:00:35] Final note: the root (".") will be signed in July so lying DNS servers should be dropped before. [22:00:41] 1. You need to turn off v4 & make sure you have the right DNS server (i.e. the DNS64) configured by hand [22:00:45] [22:00:58] http://www.ietf.org/proceedings/10mar/slides/behave-11.pdf [22:01:00] slide 2 [22:01:05] [22:01:16] 2. You need to run this: [22:01:17] dscacheutil -flushcache [22:01:21] [22:01:50] behcet.sarikaya leaves the room [22:01:59] jinmei leaves the room [22:02:18] jinmei joins the room [22:02:21] fdupont: do you have comments on the DNSSEC approaches outlined in the dns64 draft? [22:02:29] slide 3 [22:02:45] billvs leaves the room [22:03:37] Phliip Matthews [22:03:38] billvs joins the room [22:03:45] as for nat64@ietf with mac, I hear from Marc that it may not work for some domain names (where CNAME is involved) due to a bug of snow leopard. [22:03:46] JonathanLennox leaves the room [22:04:00] Hadriel Kaplan [22:04:17] the CNAME stuff turns out to be a problem in the way the implementation worked [22:04:44] because it didn't return the whole chain of CNAMEs. This is a problem in the specification. I've already proposed text to fix. [22:05:26] http://www.ietf.org/proceedings/10mar/slides/behave-7.pdf [22:05:55] magnus leaves the room [22:05:58] slide 2 [22:07:04] slide 3 [22:07:22] I missed the agenda bashing - did we move draft-thaler-6man-unique-v4mapped-00 to some other point, or did I miss it or what? [22:07:51] I think that is one of the pair that was moved around [22:07:54] at the end [22:07:57] slide 4 [22:08:09] thanks [22:08:23] satoru.matsushima joins the room [22:10:20] slide 5 [22:11:10] slide 6 [22:11:52] slide 7 [22:12:05] Dan Wing, no hat [22:12:11] return to slide 3 [22:12:15] Colman Ho leaves the room [22:13:12] behcet.sarikaya joins the room [22:15:43] Tony Hain [22:17:23] [name?] [22:20:21] andrew sullivan [22:20:25] alain durand [22:20:42] Mark Andrews [22:20:48] magnus joins the room [22:21:11] I don't want to belabour this point at the mic [22:21:26] But yes, if you're a business it's better to give your customers a non-crappy service [22:21:34] that might in fact be a reason to avoid NAT64 [22:21:47] or to overprovision so that it's adquate service anyway [22:22:14] or whatever. We need an argument for this claim that it's obviously better if the v4-initiated traffic comes in on v4 [22:22:33] particularly given the point Dan made about all those NAT44s being in the way anyway. [22:22:39] [22:22:45] Remy Despres [22:23:04] (spelling?) [22:23:29] <محسن> Rémi Després [22:23:41] Currently speaking: Mohamed Boucadair [22:23:47] (well, just finished..) [22:23:57] I think this is Mohamed. ... oh, what Dan said [22:24:28] محسن [22:24:28] : [22:24:31] thanks! [22:24:47] <محسن> u're welcome :-) Andrew [22:25:18] http://www.ietf.org/proceedings/10mar/slides/behave-2.pdf [22:25:20] slide 2 [22:25:52] slide 3 [22:26:36] slide 4 [22:27:35] akira.nkgw joins the room [22:28:53] slide 5 [22:29:41] slide 6 [22:30:27] slide 7 [22:31:36] Philip Matthews [22:33:24] brian.bnsmith joins the room [22:34:15] tskj leaves the room [22:34:44] healthyao2000 joins the room [22:36:16] next http://www.ietf.org/proceedings/10mar/slides/behave-5.pdf [22:36:24] slide 2 [22:36:53] slide 3 [22:38:05] slide 5 [22:38:55] jiro-y joins the room [22:38:56] slide 6 [22:39:34] slide 7 [22:40:08] slide 8 [22:40:28] slide 9 [22:40:50] Alain Durand [22:43:07] that was Fen Cao [22:43:17] Fred Templin [22:44:05] Dave Thaler no hat [22:44:12] picture on slide is slide 6 [22:44:42] Shane Amante leaves the room [22:45:48] Brian Carpenter [22:47:30] [didn't get name & couldn't read badge] [22:47:33] Dave Thaler again [22:47:37] no hat [22:48:07] g.e.montenegro joins the room [22:48:20] Alain Durand again [22:49:03] Dave again, no hat [22:49:15] 2758746fb9640fab leaves the room [22:49:40] 2758746fb9640fab joins the room [22:50:00] Shane Amante joins the room [22:50:10] http://www.ietf.org/proceedings/10mar/slides/behave-4.pdf [22:50:36] slide 2 [22:50:46] behcet.sarikaya leaves the room [22:51:10] slide 3 [22:51:19] Shane Amante leaves the room [22:52:12] cbyrne leaves the room [22:52:43] slide 4 [22:53:11] Shane Amante joins the room [22:53:18] Lars leaves the room [22:53:51] slide 5 [22:54:35] sbreynolds1970 joins the room [22:55:19] slide 6 [22:55:57] [someone [22:56:09] ] thinks it needs solving [22:56:23] someone = Hui Deng [22:56:35] thanks [22:57:45] velt leaves the room [22:59:20] Agreed [22:59:22] Andrew Sullivan [22:59:26] Brian Carpenter [23:00:08] http://www.ietf.org/proceedings/10mar/slides/behave-8.pdf [23:00:35] slide 2 [23:00:36] speaker=Shin Miyakawa [23:01:57] slide 3 [23:02:36] slide 4 [23:02:57] slide 5 [23:03:52] slide 6 [23:03:55] slide 7, sorry [23:04:52] slide 8 [23:06:09] skudou leaves the room [23:06:13] Dave reminds of the agenda tomorrow [23:06:21] Ronald van der Pol leaves the room [23:06:33] g.e.montenegro leaves the room [23:06:35] shinmiyakawa leaves the room [23:06:40] Suz leaves the room [23:06:40] Paul Selkirk leaves the room [23:06:46] donley.chris leaves the room [23:06:48] ikuhei leaves the room [23:06:55] sbreynolds1970 leaves the room [23:06:57] Shane Amante leaves the room [23:07:20] fdupont leaves the room: Computer went to sleep [23:07:20] jinmei leaves the room [23:07:29] Andrew leaves the room [23:07:43] Dan Wing leaves the room [23:07:45] tsavo_work@jabber.org/Meebo leaves the room [23:07:45] arifumi leaves the room [23:07:51] chris liljenstolpe leaves the room [23:08:28] Ashida leaves the room [23:08:58] becarpenter leaves the room [23:09:08] Gregory leaves the room [23:09:15] jiro-y leaves the room [23:12:15] 2758746fb9640fab leaves the room [23:13:16] billvs leaves the room [23:14:18] satoru.matsushima leaves the room [23:15:18] Hiromi Morita leaves the room [23:16:32] محسن leaves the room: Computer went to sleep [23:17:40] arifumi joins the room [23:21:15] healthyao2000 leaves the room [23:21:15] WesG leaves the room [23:24:29] 2758746fb9640fab joins the room [23:25:42] 2758746fb9640fab leaves the room [23:28:00] Marc Blanchet leaves the room [23:32:43] Marc Blanchet joins the room [23:35:33] Marc Blanchet leaves the room [23:36:09] Marc Blanchet joins the room [23:36:28] arifumi leaves the room [23:38:51] Marc Blanchet leaves the room [23:43:19] Marc Blanchet joins the room [23:46:03] jinmei joins the room [23:46:38] jinmei leaves the room [23:53:56] brian.bnsmith leaves the room