[15:39:30] hyperfeng joins the room [15:39:38] hyperfeng leaves the room [18:41:59] hyperfeng joins the room [19:58:31] YG joins the room [20:00:16] maciekwojciechowski joins the room [20:00:39] Ted Hardie joins the room [20:00:43] jon-ietf joins the room [20:00:54] martin.thomson joins the room [20:00:55] Is there anyone non-local in the jabber room? [20:01:02] Yep [20:01:15] Robert Raszuk joins the room [20:01:25] If you want one of us to forward a comment to the mic, please preface it with MIC: [20:01:25] John Morris joins the room [20:01:44] Great. [20:01:45] momose joins the room [20:01:59] Are you able to hear the audio stream? [20:02:07] Yes, everything's fine. [20:02:15] sandoche2k joins the room [20:02:15] Okay. [20:02:21] Now on the agenda slide [20:02:55] JonathanLennox joins the room [20:02:55] jlcJohn joins the room [20:02:56] There will be an overflow session 16:10-17:00 in Continental 7 & 8 [20:03:14] Those will probably be audiocast as well, but it may require some re-jiggering to find the audio stream. [20:04:33] resnick joins the room [20:04:38] ruriham joins the room [20:04:47] vidyan joins the room [20:04:54] Eric Burger now speaking [20:05:13] i'm not sure there will be audio streaming of the overflow session [20:05:20] unfortunately [20:05:31] :/ [20:06:13] jon: Are these not in use for other IETF sessions? [20:06:26] Usually an mic'ed room leaves the mics live. [20:06:40] the room is actually the iesg room i believe [20:06:49] that's all we could get for the time slot [20:07:04] afaik there's no mic [20:07:42] Pete Resnick speaking [20:08:15] Jon: okay. Possibly we can use skype or something similar to handle it. [20:08:43] I am unable to do it, as I have a conflict in that schedule slot. [20:09:57] shamus joins the room [20:09:59] Now Sebastian Kiesel [20:10:05] i'll see what i can do, but it may be tough [20:10:23] I can donate a skype account ;) [20:10:30] heh [20:10:51] Lisa Dusseault joins the room [20:10:56] Martin joins the room [20:11:34] Now on changes since -01 slide [20:11:50] Now on Table of Contents slide [20:12:03] Terminology and framework slide [20:12:16] wang.russell77 joins the room [20:13:00] The framework and arch on the slides appears in terms of peers, but still *so* much of the document talks in terms of clients and servers. [20:13:04] Question to the group: is the illustration useful. [20:13:15] (it is non-normative) [20:13:26] mlepinski joins the room [20:14:09] Barry Leiba joins the room [20:14:19] enrico joins the room [20:14:38] Question to the AD: will Alto work on one or more than one protocol? [20:14:55] eburger joins the room [20:15:05] "We want to do the right thing; generally, working on more than one likely to require careful thought>" [20:15:42] thanks, that's better than I worded it [20:15:59] Now on Requirements: ALTO Client Protocol [20:16:29] Now on new slide with the same title: "Error handling" is the first bullet [20:16:45] (Hmm, slide numbers are/would be nice, eh?) [20:17:29] Question was: must use TCP for errors, or for all traffic? [20:17:49] DCCP is out? [20:17:57] Atarashi Yoshifumi joins the room [20:17:58] Disagrees with RQv02-18 [20:18:10] Martin at the mic, replying tot he previosu speaker [20:18:32] tot he-->to the [20:18:58] Question on TCP handshake burden going onto the mailing list. [20:19:17] Requirements: ALTO discovery [20:19:37] side note: v02-23 seems very vague/silly to me. [20:20:18] Now on Requirements: Security [20:20:39] Host Location Attributes is the new slide [20:21:21] gigix73 joins the room [20:21:42] and yeah ted, there are some reqs that are a bit obvious or obscure [20:21:53] Bernie joins the room [20:23:22] Ted says: We'll need more over time. e.g., AS might have large number of other things in the routing system (ISIS areas, etc.). Can't determine that this is exhaustive. [20:23:48] Ted Q: Is the protocol for resolving group ID in the ALTO protocol? [20:24:50] wang.russell77 leaves the room [20:25:02] (I'll let Ted type the conclusion.) [20:25:18] I suggested that we not try to capture any mapping protocol requirements in this document. [20:26:46] eburger leaves the room [20:27:19] Now on the conclusion slide [20:28:38] Lisa says that locking in now probably premature. [20:29:39] wang.russell77 joins the room [20:29:43] did Lisa just say that we need the ability to change the requirements to fit the solution once we have the latter or did I misunderstand? [20:29:44] Lee Howard joins the room [20:29:50] mlepinski leaves the room: Replaced by new connection [20:29:50] mlepinski joins the room [20:30:20] No, I think she's saying is the requirements may evolve during the discussion, especially since it has presumptions of what the trade-offs are. [20:30:33] You find out new data about the trade-offs as you go. [20:30:45] yep [20:30:56] okay [20:31:17] particularly a requirements document that makes a fair amount of assumptions of mechanisms and choices [20:34:50] John Morris leaves the room [20:34:51] John Morris joins the room [20:35:03] Now moving to solution proposals [20:35:03] jon-ietf leaves the room: Replaced by new connection [20:35:41] I'm sorry, I arrived slightly late. Was there a talk on the problem statement draft? [20:35:45] jon-ietf joins the room [20:35:46] p4p/Info Export (Richard Yang speaking) [20:35:52] Mlepinski: yes [20:36:01] Will be a living document, adopted as a wg draft. [20:36:09] Thanks [20:36:10] Fair amount of discussion on certain aspects [20:36:14] Gonzalo joins the room [20:36:21] Now on slide 2 [20:36:43] Slide 3: Objective [20:37:35] Now on slide 4: My-Internet View [20:38:10] Now slide 5, Alto Query/Response [20:38:26] (How many milli-ekrs on this? .3?) [20:38:36] (er 300 milli-ekrs) [20:40:34] He's really quite near .5ekr [20:41:24] We're now on slide 7 [20:41:29] (sorry for the delay) [20:41:50] It pretty nicely matches with the audio delay :) [20:42:06] We're now on slide 8 [20:42:55] "View of World from 9th Ave". I've seen that a million times, but it's the first time I've seen the attribution. [20:43:22] Now on slide 9 [20:44:52] Slide 10, and Slide 11 [20:44:56] "Key Design Features" [20:45:24] richard.barnes joins the room [20:45:57] Now on 12 [20:46:28] On 13th [20:46:55] Alto response may specify a URI? Those aren't URIs, but fragments at best. [20:47:17] Now we are in examples, which does not appear to have a slide number [20:47:27] Alto Server/Client pair shown [20:47:54] Now showing server, peer selection [20:47:58] Shifted to Example 2 [20:49:04] ok, so groups in queries doesn't require circular reasoning because you get the groups from the server [20:49:08] Now on second slide past Example 2 title slide [20:49:29] Comparison slide [20:50:38] Request that the comparison slide return at the end [20:50:58] Now coming up: Proxidor service description [20:51:35] Now on first content slide; 3 proposals and implementations [20:52:14] 2nd Introduction slide "Address ALTO requirements" [20:53:04] Now on "Messaging" [20:54:31] Now on second messaging slide [20:56:22] Now on "Ranking System" [20:56:37] wang.russell77 leaves the room [20:58:58] Now he's on an example (p2p Neighbor/Peer selection) [20:59:40] Now on summary slide [21:00:36] Next Steps? [21:00:39] slide [21:00:48] richard.barnes leaves the room [21:01:06] Next steps does not seem to include identifying a common transport. [21:01:07] hmm. [21:02:33] John Schnizlein joins the room [21:03:02] Implementtaion slide deck. [21:03:10] Now on first slide ("p2p oracle" [21:03:13] ) [21:04:03] What is the "overflow" ALTO session chair is refering to ? [21:04:10] Now on "Proxidor server features" [21:04:38] richard.barnes joins the room [21:04:53] it is a bit strange to go through tutorials in the actual session and hold an overflow session during the break to have the actual discussion [21:05:22] Oh is that the plan ? [21:05:34] We're now on initial performance results. [21:05:34] yes, as far as I understood [21:05:43] Thx [21:06:40] Richard Yang at the mic [21:07:09] John Schnizlein leaves the room [21:07:30] Now Martin coming to the mic [21:07:40] Purpose h1/h2/h12 drafts [21:08:16] "Where we are" slide [21:09:01] now on "H is for Hemispheres" [21:10:02] Now on h1 model [21:10:05] Side: Overflow session with solution discussion is at 4:10pm in Continental Room 7/8 [21:10:08] Now on h2 model [21:10:41] Now on h1/h2 server/client model [21:10:50] "nobody is happy" [21:11:05] Now on H12 Model [21:11:17] Side ... Continental Room 7/8 has no audio for remote participants .... [21:11:59] This seems to have serious problems for v6, given the multi-scope, multi-interface problem (What do I send: /64? /48?) [21:12:14] Very sparse information space too. [21:14:29] ted brings his question to the mike [21:15:54] richard.barnes leaves the room [21:16:00] Martin reminds me that the ocean is large and may not need boiling [21:16:43] Fyi, I was talking about https://datatracker.ietf.org/drafts/draft-kitamura-ipv6-record-route/, not traceroute [21:16:50] now Saumitra Das, talking to his Goals slide [21:17:09] am I deaf, or has the sound quality been uniformly terrible for this session? [21:17:23] No, it's not great. [21:17:35] Someone is asking Saumitra to move the mic [21:17:50] For me the sound is just excellent .. and I am in Europe :) [21:18:11] Now on the assumptions slide [21:18:29] (it's actually not that good in the room, so I assume it is not that great elsewhere) [21:18:34] Now on Discovery [21:19:29] Now on "Configuration Document" [21:19:38] [21:20:47] Now on "common header" [21:20:50] Now on Message contents [21:21:02] Now on Location Query and Response [21:21:08] richard.barnes joins the room [21:21:09] Martin leaves the room: Replaced by new connection [21:22:03] Sorry, This font is like crawling ants (Host Location Attribute) [21:22:05] easier to compile the slide than read it [21:22:06] richard.barnes leaves the room [21:22:21] Saumitra is young, and his eyes work. [21:22:27] Now on Guidance Query [21:22:43] h12 joins the room [21:23:26] Now Guidance Query [21:24:11] John Schnizlein joins the room [21:24:20] Skipping to Guidance Response [21:24:32] richard.barnes joins the room [21:24:55] Now error response [21:25:26] Now on Example Use [21:26:19] Noting that there is synergy with Richard Yang's proposal [21:26:55] Use period between sessions to rest, reflect. [21:27:01] Now discovery [21:27:21] Alto Discovery Discovery discussion [21:27:41] Now on Agenda slide [21:29:42] Now in Discovery Metrics [21:32:16] Sorry, I am well behind here. We're on DNS slide [21:33:12] Now on multicast [21:36:16] Now on Discovery by Tracker [21:37:03] No on manual config [21:37:28] Now on concerns [21:38:49] Now on next steps [21:39:50] Back to DHCP slide [21:42:00] lellel joins the room [21:43:38] Barry Leiba leaves the room [21:47:01] Now Yunfei Zhang [21:47:06] ATTP slide [21:47:15] Now on Overview slide [21:47:53] Now on Problems [21:47:55] I'm not able to even be online during the second session, but is it possible to record the session and publish the mp3 later on? [21:48:07] When is the second session? [21:48:17] it's at 16:10 [21:48:37] in Continental Room 7/8 [21:48:41] thx [21:48:49] unfortunately (since that's a breakfast room) there's no audio rig in there [21:49:03] h12 leaves the room [21:50:04] Now on Advantages of the 2nd option [21:51:18] Now on ATTP: One Mechanism [21:51:28] Now on ATTP Process [21:52:58] shamus leaves the room [21:52:59] jon-ietf leaves the room [21:53:01] momose leaves the room [21:53:14] jon-ietf joins the room [21:53:34] Now on Strategies of Alias Tracker [21:53:35] YG leaves the room [21:53:35] JonathanLennox leaves the room [21:53:42] h12 joins the room [21:54:43] momose joins the room [21:54:49] Now on conclusion slide [21:54:57] ruriham leaves the room [21:56:06] h12 leaves the room [21:56:30] wang.russell77 joins the room [21:57:23] sandoche2k leaves the room [21:57:29] we're done [21:57:30] Ted Hardie leaves the room [21:57:30] momose leaves the room [21:57:35] Thanks Ted! [21:57:40] enrico leaves the room [21:57:42] jon-ietf leaves the room [21:57:44] gigix73 leaves the room [21:57:45] Lee Howard leaves the room [21:57:49] Atarashi Yoshifumi leaves the room [21:57:51] maciekwojciechowski leaves the room [21:58:03] wang.russell77 leaves the room [21:58:14] Would the jabber work at the overflow session ? [21:58:20] Thx Ted !!! [21:58:42] jlcJohn leaves the room [21:58:56] Gonzalo leaves the room [21:59:04] martin.thomson leaves the room [21:59:05] Lisa Dusseault leaves the room [21:59:19] richard.barnes leaves the room [21:59:24] Lisa Dusseault joins the room [21:59:37] Lisa Dusseault leaves the room [22:03:36] resnick leaves the room [22:04:02] vidyan leaves the room [22:05:02] richard.barnes joins the room [22:06:15] John Morris leaves the room [22:06:22] h12 joins the room [22:06:30] richard.barnes leaves the room [22:07:35] John Schnizlein leaves the room [22:11:34] jlcJohn joins the room [22:11:38] jlcJohn leaves the room [22:13:06] mlepinski leaves the room: Replaced by new connection [22:15:41] Bernie leaves the room: Replaced by new connection [22:15:41] Bernie joins the room [22:17:43] Bernie leaves the room [22:33:55] Lisa Dusseault joins the room [22:43:31] Lisa Dusseault leaves the room [22:53:10] hyperfeng leaves the room [22:57:05] lellel leaves the room [23:04:37] h12 leaves the room [23:09:22] John Morris joins the room [23:09:31] John Morris leaves the room [23:13:36] sandoche2k joins the room [23:15:44] sebastian.kiesel joins the room [23:20:04] h12 joins the room [23:20:14] h12 leaves the room [23:20:20] h12 joins the room [23:20:21] h12 leaves the room [23:20:45] Martin Stiemerling joins the room [23:21:24] this is sebastian. my laptop crashed ... [23:21:46] vijay: goal of this session is to find whether we can agree on one protocol [23:21:58] there is infoexport preferred by p2p users [23:22:08] oracle preferred by operators [23:22:16] h1h2 tries to bridge the gap [23:23:02] jon: warning from history: some time ago folks thought IETF should do chat ... now we have several standardized protocols as agreeing on one was difficult [23:23:48] anja: should keep in mind that there are other apps as p2p which can benefit from alto [23:24:47] sebastian.kiesel leaves the room [23:25:01] stas: currently isp only can know which peers you actually connect to [23:25:27] p2p client knows much more potential peers it doesnt connect to - dont want to disclose them [23:26:23] anja: the ranking in our original proposal was to disclose as few info as possible [23:27:46] stas if you ask the oracle enough questions, can you find out all relevant info [23:28:19] martin thats irrelevant - why bother asking so many question - can reverse engineer all that from bgp looking glass [23:28:58] stas: isp would be more confident if there was a FILE which will be disclosed - exactly that [23:29:52] anja: want to publish different info - difficult with a static file [23:30:46] jon: what will be the part in the middle where h1h2 is trying to solve [23:31:24] martin: i think we can build a protocol to give good guidance [23:31:39] sebastian.kiesel joins the room [23:32:39] ??? : if no isp wants to participate in ellipse 1 and no p2p wants to participate in ellipse 2 we are not very far [23:33:01] we need a protocol that covers ellipse 3 [23:34:50] jon: does anybody object agains building a flexible protocol that covers all [23:35:07] stas: protocol must clearly be incapable of snooping on users [23:36:09] ? (bruce?) : there are different kinds of apps and users - not only bittorrent [23:36:25] jon: who really feels that we should exclude one or the other model? [23:36:34] (nobody) [23:36:46] stas: i have concerns ... dont want to repeat them [23:37:54] jon: if protocol had the mechanisms to disclose from p2p to operator, but the implementor could limit them, would you use it? [23:38:07] stas: the biggest problem is the perception of the users ... [23:38:35] jon: lots of protocols exist for which the press or blogs publish strange statements [23:38:45] Martin Stiemerling leaves the room [23:39:26] ?? : were trying to draw the bar where pricacy should be ... [23:39:54] ... but different useres might make different choices ... therefore we must allow users to choose [23:40:27] stas: OUR users are fine with the performance they get ... its the ISPs who are unhappy with the situation - THEY want that something changes! [23:41:12] anja: in the end, users will benefit from better efficiency .. [23:41:53] martin: there is an ISP, which gives flat rate within own network and at night, but transit traffic during daytime is limited to 10G [23:42:04] stas: how do find out what is local? [23:42:11] anja: querying ALTO ? [23:42:21] stas: ranking alone wont help [23:43:28] richard w. if several apps could chose different models, would that stop paranoia? [23:44:03] bruce: different pricing could be an incentive for users to disclose more than now [23:46:05] anja: the figure is a little bit confusing ... ellipse 1 (rating) which is closer to "p2p related" is what the p2p users DONT want but the operators want [23:47:07] richard b: p2p users will actually benefit from more efficient transit [23:47:45] not every p2p transaction should have to start learning [23:48:41] richard y.: we have made study ... option 1 will not give better network privacy [23:49:16] optiin 1 (sorting oracle) also has severe scalability isues [23:49:49] anja: we can produce other study that says option 1 is good for operator privacy [23:50:09] richard y: we support caching [23:50:24] anja: bittorrent is just one possible app [23:51:03] jon: we need to figure out what our protocol should do .. then we can see whether one of the proposals already does this [23:51:53] richard y: my p4p proto also supports ranking - need to document this usecase [23:52:09] jon: anybody opposed to allowing the superset of all proposals? [23:52:39] Robert Raszuk leaves the room [23:53:37] ??? is there consensus that we should build a protocol that supports the full range [23:54:21] loud HUM for working on such a protocol, very low HUM opposed [23:55:05] no consensus on whether we need a negotiation mechanism [23:55:52] presentation on DNS based IP NetLocation Service [23:56:07] presenter form china telecom [23:59:52] proposal: use DNS and domain .nl.arpa. for netlocation service ...