[00:55:35] fisher joins the room [00:55:49] fisher leaves the room [00:56:03] Roger Wang joins the room [00:56:03] rbonica joins the room [00:56:18] sm joins the room [00:57:20] Colman Ho joins the room [00:57:26] jinmei joins the room [00:58:42] behcet.sarikaya joins the room [00:58:48] joel jaeggli joins the room [00:59:21] anyone listening to the streaming yet? [00:59:34] is it ok? [00:59:38] thanks [01:00:13] agenda being presented [01:00:16] meeting twice [01:00:25] 11 discussions on the agenda [01:00:30] 11 more on friday [01:00:49] fred - this is a fast paced meeting, 13 minutes per draft [01:00:58] Streaming works [01:01:02] we want to encourage discussion [01:01:52] fred - agenda is bashed [01:01:58] euijong hwang joins the room [01:02:08] dan wing is first presenter [01:02:19] behcet.sarikaya leaves the room [01:02:19] draft - happy eyeballs [01:02:28] Dave Thaler joins the room [01:02:32] keep users happy with ipv6 so they don't turn it off [01:03:03] Atarashi Yoshifumi joins the room [01:03:06] draft proposes some improvements to application behavior [01:03:39] sometimes v4 responsiveness is faster for ipv4 [01:03:55] send probes to learn which paths actually work [01:04:11] and when it breaks recover quickly [01:04:40] that fact that this doesn't work drives v6 white listing and blacklisting and so forth [01:05:09] if application performance matters like web browsers use two threads [01:05:25] anthony.baire joins the room [01:05:43] have this implemented in the links browser [01:05:51] quick fallback needs work [01:06:11] depedant on underlying interfaces that the host has. [01:06:43] applications that car about quick responsiveness we would recomend implement this [01:06:52] questions: [01:06:57] ? [01:07:09] remi depres [01:07:15] fully support idea [01:07:28] ? - optimize around extra syns? [01:07:55] dan we have some ideas to reduce rate [01:08:46] shinmiyakawa joins the room [01:08:56] sheng jian - question [01:09:10] dan - we have some work that should probably go to tsv [01:09:33] brian carpenter - referal problem apps people would love to solve this [01:10:02] fisher joins the room [01:10:08] ? - the majority of internet users think the internet is the web and and their expirence is important [01:10:31] clin perkins I support this pacing is important, to many connections at once can be a problem [01:11:15] fred - if you use sctp on the right version of linux all connections are fired at once. [01:11:34] possible concerns about multiple probes through stateful devices [01:11:49] this other protocol called tcp does this all the time [01:13:07] dan - the purpose of the draft is to make the delay muck faster than tcptimeout encourage applications to have a timer. [01:13:16] we solved that in routing sort of [01:13:24] fred - thanks dan [01:14:16] fred - we're going to do a number of polls that I encourage people to participate in (in survey monkey) was sent to the mailing list [01:14:54] please look at it over the next week, consensus if determined from the polls would be verfied on the lists. [01:15:42] fred - presented in tsv area issues related to happy eyeyballs [01:15:55] tcp reaction to soft errors [01:16:09] opening tcp connectins in complex environments [01:16:16] testing eyeball happyness [01:16:24] becarpenter joins the room [01:16:40] started discussing with benchmark methodology people. [01:16:52] if tsv area has a better idea we're all ears. [01:17:16] present ole troan [01:17:34] ipv6 mhmp problem statement [01:17:49] Mohsen joins the room [01:18:13] multihomed to non-congruent networks, could also be a proxy for being connecting to v4 and v6 [01:18:42] slide 4 [01:19:10] mif installation of static routes given multiple routes [01:19:17] and dns selection policy [01:20:57] work is going on in mif and 6man [01:22:01] Tina TSOU joins the room [01:22:08] ole troan - next presentation [01:22:23] advanced requirements for ipv6 cpe devices [01:22:57] dudisaki joins the room [01:23:16] added 6rd [01:23:20] added ds light [01:24:02] shep joins the room [01:24:31] what do you think a home topology is [01:25:12] chained ipv6 cpe's in v4 this results in multilevel nat [01:25:21] in the current cpe draft this doesn't work [01:28:41] danwing joins the room [01:30:23] we don't have a written down solution to lots of the problems in the advanced/cascaded cpe space which is a problem from the referential draft. [01:31:08] tom herbst - zigbee 802.15.4 presentation [01:31:35] in ca alone there are 5.5. million electric meteres. [01:31:53] 15k meters per day [01:32:34] ip on 51.4 only v6 not v4 adaptation layer [01:32:46] ip apckets on a 128 byte mtu [01:33:04] ripple protocol route over [01:33:13] make an ip mesh look like a subnet [01:33:28] ekline@jabber.org joins the room [01:33:28] neighbor disovering is very different than ethernet [01:34:22] multiple routers likely [01:34:58] a two router home for the basic functions that the home users wants [01:35:34] focused on ulas for the haw, you don't wnat ot be in a situation where you can't get across your home due to your isp being down [01:35:57] multiple ula problem - can't think of a way to avoid [01:36:17] multiple simultanious ulas [01:36:33] once you have two routers you're going to need a routing protocol [01:36:45] leave as an exercise for the user [01:36:59] magic packet, dhcp option [01:37:08] for firewall configuration [01:37:16] all sorts of security problems [01:37:28] use mdns for the discovery meachanism [01:37:36] only defined for link local [01:37:48] not useful for a mesh [01:38:01] need site local scope [01:38:22] next steps mdns requireements to cheshire [01:38:51] kurtis - advance mdns in dnsext [01:38:54] Suz joins the room [01:38:57] tom yes [01:39:18] ole troan what does the room think? [01:39:50] jari arko - requirements for connectivity [01:40:18] do you want to connect to it remotely from beyond the home [01:40:38] device should be able to accept a global address but not required [01:41:11] bhoeneis joins the room [01:41:24] dave thaler - express a prefer from something closer to 1 eg limited scope to single router netowork [01:41:40] rather than routed home topology [01:41:53] did you consider nd proxy [01:42:21] the issues you brough up have huge disadvantages regarding getting consensus [01:43:17] the bottom box in the chained case consider the bottom device as a bridge [01:43:33] tom - the quick anser is layer violation [01:44:22] thaler site local multicast and routing protocol select are huge issues [01:44:42] tom - you're asking for for translation bridges [01:45:11] ? go all the way to number 3 solve the whole routed homes set of problems [01:45:29] different layer2s cannot be brdiged easily [01:46:21] we can agree that the home network realy needs to evolve from where we were in the 90s [01:46:37] mark townsley [01:47:05] lee howard - the current cpe route draft doesn't break anything we see down the road [01:47:13] for this scope we need to slove this problem [01:47:18] Andrew McGregor joins the room [01:47:30] this is getting more completicated solving incemental problems will bite us [01:47:45] How can the home network determine the administrative boundary? In a chain of CPEs how does one of them know it's the "provider edge"? [01:47:56] tony hain - we have to go all the way to 3 I disagree [01:48:09] with dave thaler [01:48:13] behcet joins the room [01:48:20] mark was trying to get there [01:48:27] Andrew McGregor leaves the room [01:48:52] =) [01:48:52] ole - ekline the magic isp bit [01:48:56] samante joins the room [01:49:15] remi dpress I support item 3 [01:49:49] provide a short term solution for 1 and 2 [01:49:53] haruhikonishida joins the room [01:50:09] Andrew McGregor joins the room [01:50:23] kurtis - one and three aren't mutually exclusive [01:50:51] ole - take the basic ipv6 ce router doucment and do one and then ahave another track [01:51:15] fred templin - if the home netowork has a pi prefix this doesn't matter [01:51:54] I have trouble to hear certain speakers. is the mic on? [01:51:56] ? - support number 3 wifi and zigbee collision will occur here shortly [01:52:28] mark t - we need to go beyond 3 . 4 is real protocol work [01:52:56] fred b - something needs to be specified [01:53:05] it could already exist. [01:53:53] dave thaler - recalling homegate the dicussion of the home- toplogy included both v4 and v6 [01:54:05] fdupont joins the room [01:54:30] jari arko - requirements document that explains how we can do this with things we have. [01:54:48] kurtis dependance on procolas that might yet not be there. [01:55:02] beyond one hop dns doiscovery for example - jari arko [01:55:26] yeah the mics are on [01:55:48] fred - going back to comments on charter, we are in fact chartered to rpoduce requirements [01:55:59] Ron Bonica agrees with Sarah Palin [01:56:07] and Fred [01:56:18] don strek - pge we have a collection of utilities that can write the requirements for you fairly quickly [01:56:28] [준형] joins the room [01:56:36] nmm joins the room [01:56:54] tony hain - what is in scope for this charter, we're talking about new topologies [01:57:04] the requirements are in scope [01:57:15] the technologies are out of scope [01:57:36] fred - trolling [01:58:16] ole - in parallel we can do 1 and 3 [01:58:23] cbyrne joins the room [01:58:42] fred - what you just said is basically what we hear. [01:58:54] back on schedule [01:59:31] gunter van de velde - presents network signalling for ipv 44 /ipv6 protocol selection [01:59:41] for end systems [01:59:54] from the netowrk operator perspective [02:00:42] draft objective - provide a problem statment for the network controlled ipv4 vs ipv6 protocol preference. [02:03:32] disrupts chinecken and egg problem because you can deterministically ensure which hosts are using it [02:04:02] Dave Thaler leaves the room [02:07:01] potential overlap - draft fjisaki-6man-addr-select-opt [02:07:28] draft-ietf-6man-addr-slect-consderations [02:07:29] tatsuji.ue joins the room [02:08:18] danwing leaves the room [02:08:24] brian carpenter - draft is not just a problem statement [02:08:31] also ahs requirements [02:08:47] another is that a host can ignore this [02:08:52] gunter - yeah agree [02:09:17] dave thaler this is a good idea [02:09:39] do yo uhave a sense if merging with 6man doc is a good idea [02:09:41] danwing joins the room [02:09:45] tatsuji.ue leaves the room [02:09:52] tatsuji.ue joins the room [02:10:10] things that 3484 didn't solve [02:10:47] colin perkis - there's an aweful of equiment that hasn't been upgradeed and wil ltherefore ignore it. [02:11:36] tatsuji.ue is now known as Ue [02:11:50] fred - could you solve this with happyeyeballs [02:12:03] danwing leaves the room [02:13:09] stig - solutions today where you can prefer v4 and v6 [02:14:33] joel: agreed. you could even announce RAs w/o the autoconf bit, and only hosts that manually configured an address would be able to use v6. [02:14:36] fred wheterh or not to adopt it is in survey moneky [02:14:48] non-managed tunnels considered harmful [02:15:02] niddled to death by ducks on last version of draft [02:15:42] tsavo_work@jabber.org/Meebo joins the room [02:16:28] defined a managed tunnel if connectivity is not working someone can fix it under an sla [02:16:44] contactable administration realm [02:17:12] don't always have control of the return traffic path [02:18:32] ywang830 joins the room [02:19:01] bje joins the room [02:20:01] bje leaves the room [02:20:41] tunnel experiences - the end user view is pretty bad [02:20:58] danwing joins the room [02:21:23] content providers see impact on rtt and avoid using them [02:22:39] why do non-managed tunnels exist? [02:22:43] early adopters [02:22:56] becarpenter leaves the room [02:23:07] decouples infrascruture ipv6 readiness and deployment [02:23:25] anycast well kown address drive asymetirc connectivity [02:23:37] becarpenter joins the room [02:24:58] cannot provide a commercial sla on this service [02:26:22] up till now people would be quite happy to use these things [02:27:17] whitelist complexity [02:27:33] comments? [02:27:58] wes george - I would have less of a problem with this draft if it covered what's hapepning now [02:28:10] e.g. new local 6to4 relays [02:28:31] unmanaged tunnels are harmful, so do things to make it less harmful [02:29:04] drop boxes in your own network so at least one direction works better. [02:29:17] ywang830 leaves the room [02:29:52] brian carpenter - the toothpaste cannot be put back in the tube [02:30:26] you cannot get rid of 6to4 [02:30:31] it's here to say [02:31:07] yule ? - we don't really control as operators. [02:31:22] john mastracky - we dpeloyed 6to4 gateways [02:31:31] (comcast) [02:32:35] the thing sthe we found were problematic - were people with broken home netowkrs not so much for return path [02:33:11] dave thaler - you did a good job, the controversy meter is fairly accurate [02:33:21] the case made is valuable [02:33:22] anthony.baire leaves the room [02:33:42] anthony.baire joins the room [02:34:45] remi depres - the problems are explained [02:34:56] solution should avoid steps backward [02:36:04] mark townsley - happy eyeballs would make this a lot less painful [02:36:29] fred templin - considerd harmful has negative harmful [02:37:37] 6to4 provider managed tunnels [02:37:57] victor k [02:39:00] devices in network we don't control we have to use them as well as possible [02:39:08] with their existing tools [02:39:29] Andrew McGregor leaves the room [02:39:56] considered 6to4 with nat66 [02:40:26] using longer than 2002::/16 routers [02:40:33] rahter negative commentary [02:40:47] kbransom joins the room [02:41:21] like any nat ther are going to be some breakages [02:42:26] wmhaddad joins the room [02:42:45] ignoring this group of customers is not really an option for this customers. [02:43:06] wmhaddad leaves the room [02:43:07] danwing leaves the room [02:43:13] frodek joins the room [02:43:55] wes george - don't support it as a general solution [02:44:12] make it suck less by actually making it suck less [02:44:36] brian carptern I think this is a terrible idea [02:45:02] ywang830 joins the room [02:45:04] a constructive approach is to get operators to deploymemore 6to4 [02:45:21] julio joins the room [02:45:34] Warren Harrop joins the room [02:45:38] danwing joins the room [02:45:40] remi depres - doesn't preserve to current operation of 6to4 [02:46:48] stateless nat66 not fully realize [02:47:16] Lots of operating systems are starting to depref 6to4 when rfc1918 is present (revised 3484), so will this really address anything other than traffic to IPv6-only destinations? [02:47:57] samante leaves the room [02:48:44] fred templin - when you do the translation is that something that the provider advertises [02:49:00] vickor yeah a provider block is used for for nat. [02:49:26] samante joins the room [02:49:31] Dave Thaler joins the room [02:49:43] victor - correct [02:49:46] Dave Thaler has set the subject to: v6ops -- IETF79 [02:50:44] mark townsley - if you ask us to do this and give us enough money we'll do this. don;t ask ask su [02:51:25] vincent-afnic joins the room [02:51:42] vitor - you're right it's starting go away with address selection being fixed. [02:52:35] mosion suicie - don't like this provides providers an out. [02:54:02] tim shepeherd - reading the draft, slowly came to the realziation how incredibly harmful this is [02:54:04] @joel: Mohsen Souissi :-) [02:54:15] hey close [02:54:18] heh [02:54:20] thanks [02:55:49] My point was: this proposal gives ISP a wrong and bad signal. They might think they can still buy time playing with IPv6. If they don't have 6rd, DS-lite or tunnel-broker-like mechanisms, just tell them they will encounter pain [02:56:36] Chris Griffiths joins the room [02:57:57] Suz leaves the room [02:58:06] tony this is fundamentally flawed because plain and simpel it does nat66 [02:58:54] I appreciate the concern about not wanting to break things [02:59:25] wmhaddad joins the room [02:59:44] victor - we wanted to run a test and seewhat this does and report back in prague. [03:00:13] jason livinggood - dns whitelisting - go through quickly [03:00:18] i think the scope of the problem is too small to be worth much effort. IPv6 only content? return path latency? Neither of those things are worth this much time. [03:00:56] want that read out when the line forms? [03:01:28] no, that was for PMT [03:01:46] ah ok [03:01:49] thanks [03:02:12] why are some considering it - performance [03:02:47] downsides and risks operational costs, being done on a adhoc basis [03:03:49] policy for controlling whitelists are opaque and hard to scale. [03:04:04] monitoring an troublshooting potentially challengeing [03:06:42] email if you'd like to do a detailed review of the document [03:07:13] moshen - I read it, I'm little frustrated with the solution. [03:07:14] danwing leaves the room [03:07:29] don't do it it will be harmwful and balaknize the internet [03:07:40] wes goerge it belongs in v6ops [03:08:03] so does Ron [03:08:15] there is another draft from cmpacst about how you notify sers [03:09:18] My point: the draft should express some recommendation on solutions enumerated at the end. v6ops shoud give clear recommendations. and mine is: don't do it it will harm the Internet on the long terme (balkanization of the Internet) [03:09:41] danwing joins the room [03:09:48] And of cours, it's a very nice document: should be adopted by either dnsop or v6ops [03:09:55] dany belongs here, it's important [03:10:10] the namespace presents a problem. [03:10:12] i agree it belongs in v6ops [03:10:39] fragementing dns as a policy is a general problem [03:10:50] not just in v6 [03:10:57] - dany m [03:11:23] mark townsley - happy eyeballs would help here. [03:11:36] google please give us ipv6 [03:12:29] but you know we might not have it without this [03:13:03] danwing leaves the room [03:13:07] warren one of the folk doing this does want to do away with it. [03:13:09] +1 whitelist is better than nothing [03:13:23] wej joins the room [03:14:03] tsavo_work@jabber.org/Meebo leaves the room [03:14:11] fred - survey monkey check it as a working group draft [03:14:35] bechet sakariya - ipv6 in mobile networks [03:15:19] Chris Griffiths leaves the room [03:15:30] 02 - is out of comments recived from mastricht meeting [03:16:33] [준형] leaves the room [03:17:03] ekline@jabber.org leaves the room [03:17:12] Andrew McGregor joins the room [03:17:15] Chris Griffiths joins the room [03:18:49] Warren Harrop leaves the room [03:19:31] ready for wg addoption [03:20:03] bhoeneis leaves the room [03:20:12] ? [03:20:32] joni kornen - few questions [03:20:55] wmhaddad leaves the room [03:21:09] this is still a omoving target in 3gpp [03:21:24] are you saying that the router advertisement would contain multiple prefixes? [03:21:36] because that's definenly not the case [03:21:42] Joonhyung Lim joins the room [03:22:04] bechet - in general more than one prefix should be assigned. [03:22:19] joni -in a snes this piture is wrong [03:22:25] Tina TSOU leaves the room [03:22:32] Tina TSOU joins the room [03:23:00] for the next slide again going back to 3gpp -. are you saying the mn would configur it's own and and delegate it's prefix [03:23:45] joni - i would be willing to help [03:25:17] jonne S - meta level question what is this document targets for. [03:25:31] are you traying to document what they've done [03:25:39] Tina TSOU leaves the room [03:25:45] which is hard becuase they haven't done it yet. [03:26:14] bechet dhcpv6 pd is not so easy to use? [03:26:18] Tina TSOU joins the room [03:26:23] joni - not sure what I prefer. [03:27:38] joni k - ipv6 in he in 3gpp in evolved packet networks. [03:27:55] some release 10 stuff has been added. [03:28:40] access point name concept [03:30:43] shep leaves the room: Logged out [03:32:10] haruhikonishida leaves the room [03:32:37] release 10 has dhcpd prefix delegations [03:32:51] will drive stateful dhcp deployment [03:33:51] ongoing work [03:33:53] fdupont leaves the room: Computer went to sleep [03:34:07] fred - what should we do with it. [03:34:15] it's still on ad sponsered track [03:34:22] samante leaves the room [03:34:32] behcet leaves the room [03:34:38] joni - fine if the working groups wants to carry it it [03:34:52] jonne - individual track is just fine [03:35:01] Chris Griffiths leaves the room [03:35:16] Atarashi Yoshifumi leaves the room [03:35:34] euijong hwang leaves the room [03:35:36] nmm leaves the room [03:36:00] cbyrne leaves the room [03:36:03] Joonhyung Lim leaves the room [03:36:07] joel jaeggli leaves the room [03:36:09] ywang830 leaves the room [03:36:10] rbonica leaves the room [03:36:13] kbransom leaves the room [03:36:16] vincent-afnic leaves the room [03:36:16] Andrew McGregor leaves the room [03:36:29] julio leaves the room [03:36:33] Colman Ho leaves the room [03:36:52] fisher leaves the room [03:37:31] shinmiyakawa leaves the room [03:38:31] Roger Wang leaves the room [03:38:50] Ue leaves the room [03:42:32] Bernie joins the room [03:43:33] Dave Thaler leaves the room [03:43:35] Dave Thaler joins the room [03:43:54] anthony.baire leaves the room [03:44:25] dudisaki leaves the room [03:44:33] cbyrne joins the room [03:44:56] sm leaves the room [03:44:57] danwing joins the room [03:44:57] danwing leaves the room [03:45:03] becarpenter leaves the room [03:45:07] jinmei leaves the room [03:50:11] frodek leaves the room [03:52:59] Dave Thaler leaves the room [03:54:17] cbyrne leaves the room [03:57:26] Mohsen leaves the room: Computer went to sleep [04:01:40] danwing joins the room [04:06:07] danwing leaves the room [04:09:30] jinmei joins the room [04:12:31] Colman Ho joins the room [04:13:54] Colman Ho leaves the room [04:20:04] Tina TSOU leaves the room [04:25:12] danwing joins the room [04:26:23] danwing leaves the room [04:26:47] danwing joins the room [04:50:04] Bernie leaves the room [04:57:08] danwing leaves the room [05:02:08] Bernie joins the room [05:04:20] jinmei leaves the room [05:09:32] danwing joins the room [05:10:47] samante joins the room [05:13:45] samante leaves the room [05:17:54] samante joins the room [05:31:47] danwing leaves the room [05:35:58] danwing joins the room [05:45:59] danwing leaves the room [06:26:36] Bernie leaves the room [06:26:43] Bernie joins the room [06:31:06] Bernie leaves the room [06:32:59] samante leaves the room [06:42:12] Andrew McGregor joins the room [06:55:55] Bernie joins the room [07:07:36] Bernie leaves the room [07:18:12] Andrew McGregor leaves the room [07:19:10] anthony.baire joins the room [07:20:37] anthony.baire leaves the room [11:39:28] wej leaves the room [12:51:39] wej joins the room [18:56:24] wej leaves the room [19:01:33] wej joins the room [22:20:07] wej leaves the room