IETF
homenet@jabber.ietf.org
Wednesday, 28 March 2012< ^ >
Room Configuration

GMT+0
[04:10:30] dtaht joins the room
[05:43:45] dtaht leaves the room
[05:50:54] dtaht joins the room
[06:29:36] <dtaht> grumble
[06:56:46] tsuichi joins the room
[07:01:06] gigix73 joins the room
[07:01:30] SM joins the room
[07:01:39] bernie (ucom.ch) joins the room
[07:01:52] ogud joins the room
[07:02:14] wmtownsley joins the room
[07:02:49] Wassim M. Haddad joins the room
[07:03:06] Randall Gellens joins the room
[07:03:30] <wmtownsley> Welcome to homenet.
[07:03:42] Ralph Droms joins the room
[07:04:09] mcharlesr joins the room
[07:04:20] <mcharlesr> hi. I'm the what is going on scribe.
[07:04:20] Klensin joins the room
[07:04:50] <mcharlesr> say MIC: if you want me to channel you.
[07:05:35] <mcharlesr> slide: "updates from -01"
[07:05:39] Dave Thaler joins the room
[07:07:23] <mcharlesr> slide: "Started to use issue tracker"
[07:08:22] alex joins the room
[07:08:48] Andrew McGregor joins the room
[07:08:48] <mcharlesr> eight issues so far: http://trac.tools.ietf.org/wg/homenet/trac/report/1
[07:11:12] jpc joins the room
[07:11:50] <mcharlesr> any comments about how adverturous we want to be?
[07:13:48] Suzanne joins the room
[07:14:01] teco.boot joins the room
[07:15:01] <mcharlesr> issue #1: http://trac.tools.ietf.org/wg/homenet/trac/ticket/1
[07:17:45] <Klensin> Could someone please remind those in the room that those of us who are remote may not recognize voices and don't have visual clues -- it is _very_ hard to follow what is going on in this sort of multiparty discussions if the people aren't identified... frequently.,
[07:18:03] <mcharlesr> yes. Klensin... we are beating people.
[07:18:11] <mcharlesr> I will write who things are: Lorenzo at mic now.
[07:18:14] iljitsch joins the room
[07:18:18] <mcharlesr> (and I was first, and then Ole)
[07:18:21] <mcharlesr> Jari Arkko.
[07:18:23] <dtaht> another way to sort of solve this is to mandate that provider independent addressing is required for doing multihoming.
[07:18:39] <Andrew McGregor> Dave, you're not going to get that at home
[07:18:47] <iljitsch> not sure what Lorenzo is getting at but I don't see it working
[07:18:54] <mcharlesr> dthat: huh... residential people aren't going to get a PI/32.
[07:19:12] <iljitsch> sending packets based on source address goes counter to hop-by-hop and thus requires new code
[07:19:26] <mcharlesr> he has 6 native IPv6 uplinks...
[07:19:27] <dtaht> without nat, the failover situation is untenable.
[07:19:47] <Klensin> thanks
[07:19:47] 8bb7997aac1d8919 joins the room
[07:19:47] <teco.boot> Lorenzo is right, routers can take care of forwarding to right CER,
[07:19:50] <iljitsch> but is that so hard to implement in hosts? not nearly as hard as in routers, IMO
[07:20:24] <iljitsch> coordination between (home) routers is a problem, though.
[07:20:41] <mcharlesr> iljitsch, homenet is really about building coordination between home routers.
[07:20:57] <iljitsch> it's much better if you can get two (or more) that don't have to work together and let the host send the packets to the right one
[07:21:01] <mcharlesr> andrew mcgregor.
[07:21:28] <iljitsch> the "let the router figure it out" is much more appropriate in larger networks.
[07:21:36] kawashimam joins the room
[07:21:37] <iljitsch> (approach)
[07:21:48] <mcharlesr> Lorenzo Coletti.
[07:22:02] <mcharlesr> Dave Aren.
[07:22:20] <mcharlesr> Server Provider B "right router"
[07:22:30] tsavo_work@jabber.org/Meebo joins the room
[07:22:43] <mcharlesr> Dimitry from MS.
[07:24:17] sftcd joins the room
[07:24:29] Ole Troan joins the room
[07:24:35] bernie (ucom.ch) leaves the room
[07:24:35] <Ole Troan> Dave Thaler talking
[07:24:51] <Ole Troan> Michael Richardson
[07:25:27] bernie (ucom.ch) joins the room
[07:26:00] <mcharlesr> Lorenzo Coletti.
[07:26:22] Dave Thaler leaves the room
[07:26:23] Dave Thaler joins the room
[07:26:52] <mcharlesr> slide 2: "MH1)" and MH7/MH8/RT11/RT12.
[07:26:58] <iljitsch> I agree with Lorenzo's goal
[07:27:32] <mcharlesr> comments from floor that this model is being shipped by many.
[07:27:50] <dtaht> have people so totally written off the prospect of per city/region provider independent addressing as to not revisit it now? Dedicated address space is what customers want. All the dynamitism otherwise needed drives the end user crazy.
[07:28:03] <dtaht> and it makes multihoming and failover work.
[07:28:06] <mcharlesr> RFC3002 is "design for walled garden"
[07:28:09] alex leaves the room
[07:28:59] <Dave Thaler> To be clar: That's an IAB stream RFC, which is a workshop report, not an IAB consensus document per se.
[07:29:06] Atarashi Yoshifumi joins the room
[07:30:39] <mcharlesr> and RFC3002 is written in 2000, and I think predates much IPv6 thinking and experience with IPv4 squatting on "unallocated" addresses in "gardens". Walled Garden is better than NAT66.
[07:31:06] <Klensin> Nic: While combining all the models into one super-complex one and then doing that is a fine model, it may ignore economic reality. Remember that home boundary routers will mostly be installed in commodity situations. The Rolls Royce of home routers will be important, but the ones that cut enough corners to be low-cost commodities will get the market share. If the WG doesn't recognize that and be clear about what is minimally important, it will fail.
[07:31:21] <mcharlesr> consensus towards Dave Thaler model by hum.
[07:31:33] <Klensin> s/Nic/Mic. above
[07:31:36] <iljitsch> klensin-ietf: how do you write specs for people who don't implement specs?
[07:32:10] <iljitsch> home routers are cheap, people will buy a better one if that solves a problem that's sufficiently important to them; market forces will do the rest.
[07:32:22] <Dave Thaler> agree with Klensin (but I'm not near the mic, someone else should relay)
[07:32:52] <mcharlesr> Topic #4
[07:32:54] <iljitsch> can the blue sheets be circulated towards the back of the room?
[07:33:19] monden joins the room
[07:33:29] ws4d joins the room
[07:33:39] cabo joins the room
[07:33:59] satoru.matsushima joins the room
[07:34:14] yuichi.igarashi@gmail.com joins the room
[07:34:33] Peter Koch joins the room
[07:34:55] <Dave Thaler> @iljitsch: this wg is supposed to write specs that other sdo-like orgs (DLNA, Upnp-forum, BBF, etc) will base specs on that the cheap routers DO pay attention to. So it's indirect.
[07:35:07] gshapiro joins the room
[07:35:32] <mcharlesr> Lorenzo at mic.
[07:35:37] <mcharlesr> topic #6
[07:35:51] <mcharlesr> back to #4.
[07:36:10] <iljitsch> dthaler: right, so we should make specs that make sense, not prematurely optimize for implementation economics (although implementation considerations in general are always relevant)
[07:36:21] <mcharlesr> Jari Arko at mic.
[07:36:29] <Dave Thaler> exactly
[07:36:46] <Dave Thaler> feel free to say that at the mic when appropriate
[07:37:10] <mcharlesr> Kerry Lynn.
[07:37:21] <mcharlesr> #6: arbitrary topologies.
[07:37:31] <iljitsch> dthaler: I can't get to a mic without trampling people and hardware...
[07:37:42] <Dave Thaler> me neither
[07:37:54] <Klensin> @ iljitsch Those who are sophisticated enough to have more complex problems and know how to solve them will certainly go for higher-end, higher-cost equipment. The other 99.99+% of the community will do standard things on whatever commodity equipment is suggested by their (one) ISP or the neighborhood commodity electronics store -- and will select, recommend, and sell on price and easy configurability/maintainability by people with no knowledge whatsoever (and less patience)
[07:38:50] <mcharlesr> Klensin, so this is why folks like Jim Gettys/Dave Taht are trying to get this stuff into *WRT, such that all of this complex software is available essentially for free.
[07:39:17] dtaht blushes
[07:39:17] <mcharlesr> aC.
[07:39:20] <Ole Troan> can't we call it "prefix assignment" instead of "prefix delegation". the latter is used between separate administrative entities
[07:39:28] <dtaht> trying to get it done RIGHT... and available for free.
[07:39:38] <mcharlesr> AC.
[07:39:39] Guillaume Habault joins the room
[07:40:25] <mcharlesr> Tim Chown.
[07:40:39] <mcharlesr> Samita J...
[07:41:33] <iljitsch> klensin: it's true that obscure features will find it hard to get implemented, but if we don't believe what we're doing here provides benefits to a reasonable number of users we should all go out and enjoy the Paris scenery right now.
[07:41:38] <mcharlesr> Mark.
[07:43:31] <mcharlesr> Ralph Drums.
[07:44:36] <mcharlesr> Dave Aren
[07:44:44] <wmtownsley> *Oran
[07:44:55] <mcharlesr> (i knew i wrote it wrong)
[07:45:54] <mcharlesr> "which came first, the prefixes or the routing"
[07:45:56] <bernie (ucom.ch)> s / Drums / Droms /
[07:46:13] <dtaht> chicken/egg problems. I have.
[07:46:28] <Klensin> @mcharlesr: I didn't say "essentially for free". Actually, within broad limits, I don't think it makes a lot of difference whether the minimum-cost commodity device sells for EUR X or EUR 3*x. Acceptable installation/ setup costs have a lot more to do with expectatiosn than we normally believe. I agree with those who are talking about building in generality, both because ISPs will underestimate what they will need/want later, but I think we also have to find a balance between cost and complexity (not only technological but as perceived by users and support people) or the market will be taken over by function-restricted trash that kind of works most of the time.
[07:46:28] <mcharlesr> Jari Arkko.
[07:47:15] mcharlesr is now known as mcr
[07:47:16] mcr is now known as mcharlesr
[07:47:16] mcharlesr is now known as mcr
[07:47:59] <dtaht> http://pastebin.com/YZDnGHz7 my routing topology. which uses ahcp flooding for address assignment.
[07:48:45] <dtaht> market is already taken over by function restricted trash. How do you make a compelling argument for non-function-restricted trash.
[07:49:24] <mcr> Issue #7.
[07:49:30] <Andrew McGregor> Just doing the arbitrary topology solution sounds like the right thing. Otherwise you end up with support pain.
[07:50:11] <mcr> Eric K.
[07:50:17] <mcr> Lorenzo Coletti.
[07:50:37] mcr is now known as mcharlesr
[07:50:37] mcharlesr is now known as mcr
[07:51:28] <Klensin> @dtaht: Two answers: I think the issues raised by IPv6 --exactly the ones that are now being discussed-- creates an opportunity to get things right this time. Maybe I'm too optimistic. But, if I am wrong and the future is inevitably such trash, this WG lies somewhere on the spectrum between "second system effect" and "waste of time".
[07:52:03] <dtaht> I agree with both of your statements.
[07:52:04] <iljitsch> andrewdmcgregor: if we can support arbitrary topologies so much the better, but it's not unreasonable to draw a line somewhere once the specs and implementations would get really complex for unusual setups
[07:52:35] <iljitsch> I mean, we tell people not to connect two ports on one hub to two ports on another hub
[07:52:59] <iljitsch> that's a reasonable limitation (but we were able to lift it once we got switches and spanning tree)
[07:53:50] <dtaht> I do disagree with many bits of ipv6 history (like the portable ipv6 addresses idea being still so arbitrarily dismissed, when communicating the concept of 'ownership' of your own slice of the internet to the current crop of cyberserfs is a very powerful one)
[07:53:56] <mcr> Issue Tracker #4:
[07:54:07] <iljitsch> klensin: so we have people who think something useful can be done and one(s?) that don't, does it make sense for the latter to try to stop the former?
[07:54:19] <dtaht> heh...
[07:54:25] <wmtownsley> People speaking at the Mic: Please state name clearly, note takers having trouble.
[07:54:51] <Suzanne> thx Mark
[07:55:04] <mcr> Lorenzo Coletti.
[07:55:37] mcr is now known as mcharlesr
[07:55:37] mcharlesr is now known as mcr
[07:55:40] <Dave Thaler> I'm skeptical that you can simultaneously support the complexity of the scenarios that people in this room use at home, and still be used by the low-end gateways that take their reqts from other forums we want to influence with our work. Optimizing for the complex cases won't necessarily get those forums to adopt our stuff the way we'd want. We need to make sure we can frame things as mandatory vs optional, or something else so they don't create profiles that remove things we'd consider mandatory even for simple cases.
[07:57:32] <iljitsch> dthaler: it's ok for software to be complex if its behavior can be tested well and it provides a compelling benefit. and at the design stage you really need to look at everything. but maybe at some point pruning will be necessary, yes.
[07:57:37] <mcr> Anders Brandt.
[07:58:53] <mcr> Jari Arkko
[07:59:02] <Dave Thaler> iljitsch: those who write the software won't necessarily look at our specs. they might be looking at dlna or upnp forum or bbf or (etc) specs instead.
[07:59:14] <Klensin> @ iljitsch See Dave Thaler's notes, above (with which I agree). I am not suggesting designing for low-end trash. I am suggesting that, for "home" equipment, the economic and support realities are actually key engineering and design constraints and we must not lose sight of that if the WG's work is to be useful.
[07:59:23] <iljitsch> dthaler: hence the importance of compliance testing
[08:00:30] <iljitsch> klensin-ietf / dthaler: I think there probably is a limited but usable intersection between our viewpoints
[08:00:37] mcr is now known as mcharlesr
[08:00:37] mcharlesr is now known as mcr
[08:00:45] <dtaht> I have multiple issues with ULAs. 1) happy eyeballs is barely implemented on web browsers. (think, wget, dns, other tools) 2) the address calculation mechanism is busted. 3) dns naming has to be solved before implementing any private addressing scheme. 4) renting ips sucks in the general case (thus my advocacy of the old provider independent ip problem)
[08:00:51] <mcr> Ralph Droms
[08:00:55] <Dave Thaler> iljitsch: yes, and the compliance testing is done by those who give logos, etc... hence against DLNA etc specs
[08:00:56] <Klensin> Another element of "make this work" is to keep in mind that many networks topologies that the people in the room can set up and operate are irrelevant in the marketplace (and things that ISPs are likely to try to prohibit to minimize support costs)
[08:01:03] <dtaht> (I can live with being grumpy only about 4)\
[08:02:09] <iljitsch> klensin-ietf: people also never wanted an iPhone until they saw one. We have to be a bit more ambitious than just make a 1% better version of what sells best today, although we have to be careful to bite off too much, sure.
[08:02:30] <dtaht> iljitsch: applause
[08:03:20] <mcr> discussion Mark and Ralph.
[08:04:02] <mcr> Thom Herb(?)
[08:04:13] <wmtownsley> Herbst.
[08:04:25] Ole Troan leaves the room
[08:04:58] Ole Troan joins the room
[08:05:40] mcr is now known as mcharlesr
[08:05:58] <mcharlesr> Jari Arkko.
[08:06:15] <mcharlesr> Lorenzo Coletti.
[08:07:36] Daniel Gavelle [nxp] joins the room
[08:07:40] <Andrew McGregor> Oldest ULA should win, the one whose valid lifetime is closest to expiry
[08:07:52] <mcharlesr> Jari Arkko.
[08:08:05] gshapiro leaves the room
[08:08:22] <mcharlesr> Lorenzo Coletti.
[08:08:37] <Dave Thaler> this discussion sounds a lot like some of the zeroconf WG discussions...
[08:09:34] <dtaht> There simply aren't enough routes in a homenet to fill a routing table inside the home
[08:09:43] <dtaht> even if every router chooses a unique ULA
[08:09:51] <Suzanne> I think what they mean is "a homenet should have its own ULA."
[08:10:08] <Suzanne> and if it then wants to interact with others, that's what routing is for
[08:10:16] <dtaht> but there is no effective way currently to delegate subnets to all devices that also route.
[08:10:17] <Andrew McGregor> Multiple ULAs is not a disaster
[08:10:41] mcharlesr is now known as mcr
[08:10:41] mcr is now known as mcharlesr
[08:10:42] dtaht just throws a new ula up in the air on every router, they announce, they route, next question.
[08:10:42] <iljitsch> I'm not against ULAs but they seem to be proliferating awfully fast right now. can I still run a network without them when this wg is done?
[08:11:04] <Andrew McGregor> @iljitsch: Sure you can
[08:11:07] Chris Griffiths joins the room
[08:11:17] <iljitsch> ok then
[08:11:32] <Andrew McGregor> If it's managed, making them unique is easy, you probably actually want that.
[08:12:00] <iljitsch> don't forget that you only need ULAs for inter-subnet communication, for intra-subnet link-local is easier
[08:12:31] <Andrew McGregor> And every stack I know of will do that even in the presence of multiple ULAs and globals
[08:12:43] <Andrew McGregor> (presuming name resolution will return them all)
[08:12:43] <iljitsch> andrewd: 56 bits of randomness (add 16 for subnet) is more than good enough to not worry about collisions.
[08:12:53] <Andrew McGregor> Exacrly
[08:13:03] <Andrew McGregor> s/cr/ct/
[08:13:07] <mcharlesr> Elliot Lear. Then Michael Richardson. Then Anders Brandt.
[08:13:23] ws4d leaves the room
[08:13:41] <mcharlesr> Tim Chown.
[08:13:46] <mcharlesr> Lorenzo Coletti.
[08:14:41] <Klensin> iljitsch: If you are going to make the iPhone argument, you had better make sure that the design features you are interested in are immediately visible and obvious to the typical end-user. Today, that end-user sees the home boundary router as a mystery box that outside wiring goes into, "Internet" comes out, and that doesn't work well if it doesn't have power. To paraphrase an old friend, "people don't care about piple (or how they connect), they only care about the water or whatever else flows through them". If you can build a home router that will deliver good coffee in the mornings and soothing tea in the evenings, you wil see an iPohone effect. More sophisticated hendling of prefixes, ULAs, etc., will just earn blank looks.
[08:15:40] mcharlesr is now known as mcr
[08:15:52] <mcr> Klensin, I unplugged three "routers" in February, that were added by well-meaning people who thought they had bought a *switch* for their office. They screwed themselves buying that homenet-free device. They returned it.
[08:15:53] <8bb7997aac1d8919> (sorry, but is the agenda followed somehow? http://www.ietf.org/proceedings/83/agenda/agenda-83-homenet.txt)
[08:16:32] 8bb7997aac1d8919 leaves the room
[08:16:48] <Andrew McGregor> This may be heretical in this context, but I think a lot of this stuff is much more valuable in a small business context than in homes (although many peoples homes contain multiple small businesses…. mine has three)
[08:16:59] <mcr> if it had done homenet, and could be plugged in anyway, and it turned off dhcp correctly, then they would have kept that sale, the users in question would have been happier (The users on that subnet lost a morning of work, until someone thought to ask me) I would have been happier too.
[08:17:13] <mcr> Kerry Lynn.
[08:17:15] Alex Petrescu joins the room
[08:17:38] <mcr> #5: flash renumbering.
[08:18:35] <dtaht> I agree strongly with thinking in terms of small business context.
[08:19:53] Randall Gellens leaves the room
[08:20:08] <Klensin> @mcr: Yes. And so? Almost certainly if you can design a device that will notice when it is being installed in an improper configuration or topology, you will be able to sell it (at least to ISPs who control what gets used as CPE) on the basis of reduced support costs. To me, that is more of the basic economics of the system, not fantasies about routers as the next iPhone.
[08:20:41] mcr is now known as mcharlesr
[08:20:41] mcharlesr is now known as mcr
[08:20:48] <dtaht> and there's the persistent naming to ip mapping problem...
[08:20:56] <iljitsch> ppp? that's useless for IPv6
[08:21:06] <Andrew McGregor> Even if isp's don't control the CPE, they frequently recommend devices
[08:21:22] <Andrew McGregor> ppp works fine for ipv6, so long as you have dhcpv6 pd as well
[08:22:00] <Dave Thaler> there's this other WG called 6renum...
[08:22:06] <wmtownsley> Ole Troan
[08:22:07] <mcr> Ole.
[08:22:16] Peter Koch leaves the room: Computer went to sleep
[08:22:34] <Dave Thaler> maybe flash renumbering is out of scope for this wg to solve
[08:23:39] <wmtownsley> Dave: I think we have to consider that it is present, but agree that it might not have to work perfectly when it happens.
[08:24:07] wmtownsley leaves the room
[08:24:12] Ole Troan leaves the room
[08:24:24] <iljitsch> klensin: I'm not saying this group is on the brink of iPhone-ness, just that asking people what they want today doesn't allow for new things that people will want but can't imagine yet. As far as I'm concerned, the ability to get cable + ADSL so that when one fails you continue using the other without much if any interruption, is something useful that I believe a reasonable number of people will want. And it's good for ISPs and hardware vendors, too, because it sells more lines and more boxes. I could be wrong of course and maybe nobody wants this, but I don't see how we can know that for sure at this stage.
[08:24:58] wmtownsley joins the room
[08:25:07] <iljitsch> andrewd: that's what I mean, PPP only gives you framing for v6, not much usable parameter (= address) negotiation.
[08:25:11] <mcr> Lorenzo Coletti.
[08:25:31] Ole Troan joins the room
[08:25:42] mcr is now known as mcharlesr
[08:25:42] mcharlesr is now known as mcr
[08:25:54] <Andrew McGregor> @iljitsch: Sure, but in a practical sense ppp+dhcp6 works just fine
[08:25:55] yuichi.igarashi@gmail.com leaves the room
[08:26:13] <iljitsch> andrewd: yes, but no thanks to ppp
[08:26:17] dtaht has no love for dhcp-pd...
[08:26:24] <mcr> Suresh ..?
[08:26:49] <SM> Yes
[08:27:32] <mcr> Anders Brandt.
[08:27:54] <mcr> Thomas.
[08:28:44] <teco.boot> Can we support dynamic PD for privacy and stable PD for reachability???
[08:28:44] <mcr> next presenter: Ralph
[08:28:53] Peter Koch joins the room
[08:29:07] <mcr> "Name Service in the homenet"
[08:29:17] <Andrew McGregor> I get more stable dhcp6 ipv6 addressing than ppp ipv4 at home already… this has improved since the last ietf, where I was getting renumbered twice a day. However, *my gear didn't care about the renumbering in ipv6, where it fails horribly in ipv4*
[08:30:44] mcr is now known as mcharlesr
[08:30:44] mcharlesr is now known as mcr
[08:31:01] <iljitsch> andrewd: you had overlapping use of old and new addresses?
[08:31:18] <dtaht> dnsmasq has added dhcp4 to SLAAC naming
[08:31:21] <dtaht> commit 801ca9a7b7f26902ac7b930fa5fb4c3d91a17322
Author: Simon Kelley <simon@thekelleys.org.uk>
Date: Tue Mar 6 19:30:17 2012 +0000
Add ra-names SLAAC-hostnames from DHCPv4 option.
[08:32:53] Alex Petrescu leaves the room
[08:32:57] <wmtownsley> BTW, There is a proposal from a certain german ISP that suggests IPv4 renumbering in IPCP without a link down even in order to deal with IPv4 address exhaustion (while IPv6 remains up on the given link).
[08:33:11] <wmtownsley> s/even/event
[08:33:13] <dtaht> oh god.
[08:33:19] <wmtownsley> oh yes.
[08:33:20] <Andrew McGregor> @iljitsch: yes, for a while every time I tried to renew (every 12 hours) I got a new prefix with a valid lifetime of 24 hours, and the old one continued to work
[08:33:41] dtaht just spent 2 weeks behind a 500 port static nat as an exercise
[08:34:11] <wmtownsley> http://tools.ietf.org/id/draft-fleischhauer-ipv4-addr-saving-02.txt
[08:34:12] <dtaht> (in the future pain users will feel if we don't get ipv6 out the door)
[08:35:46] mcr is now known as mcharlesr
[08:35:46] mcharlesr is now known as mcr
[08:35:51] <mcr> light bulbs on 15.4 networks need access to printers... so you can review the EULA.
[08:36:00] <dtaht> hahah
[08:36:08] <mcr> Stuart C.
[08:36:38] <iljitsch> andrewd: nice. this is exactly how this should work, although maybe renumbering twice a day is still a bit often even if it works so well.
[08:38:44] Guillaume Habault leaves the room
[08:39:40] <mcr> Dave thaler.
[08:39:46] <mcr> that was ted Lemon before.
[08:40:47] mcr is now known as mcharlesr
[08:40:47] mcharlesr is now known as mcr
[08:40:51] <dtaht> good coverage of that problem
[08:40:51] Alex Petrescu joins the room
[08:41:14] <mcr> (I couldn't get his name)
[08:42:00] <Andrew McGregor> H Kitamura, from NEC
[08:42:35] <Dave Thaler> summary of my questions: how many naming domains, is guest network different? Is global dns namespace a requirement for guest network? Do they need names in the local naming domain or just the names they might bring in from elsewhere? What is uniqueness requirement for naming when a partition might exist?
[08:42:55] <Andrew McGregor> My view of that draft is, those names might be acceptable to Japanese users accustomed to long unmemorable strings of goop, but they're not exactly user-friendly.
[08:43:19] Guillaume Habault joins the room
[08:43:24] Thomas Heide Clausen joins the room
[08:43:37] jweil joins the room
[08:45:47] mcr is now known as mcharlesr
[08:46:16] <mcharlesr> Samita J. again.
[08:46:46] <mcharlesr> Stuart Cheshire.
[08:47:43] dtaht applauds stewart's second
[08:47:57] <dtaht> split dns, same name, different location, different resolved ip address
[08:48:30] <dtaht> s/stewart/stuart
[08:48:40] <wmtownsley> OK, stuart, thank you for two solutions. Now, pick one. I see at least one vote in here, is that the right one?
[08:48:59] <mcharlesr> Ted Lemon.
[08:49:26] <mcharlesr> homenet dns solution dealing with reverse tree.... draft-lemon-pddns-
[08:49:43] <Peter Koch> draft-lemon-dhc-dns-pd-00.txt
[08:50:48] mcharlesr is now known as mcr
[08:51:18] <Dave Thaler> is this preso about name resolution (for which we just talked about reqts) or about service discovery (where we haven't talked about reqts yet here)?
[08:51:53] <mcr> uhm. it's an extension to "mdns/bonjour", so it doesn't discover services, unless you think you do that by looking for names.
[08:52:31] <Dave Thaler> SRV records are service discovery
[08:52:38] <dtaht> heh.
[08:52:45] <Dave Thaler> in some sense
[08:52:48] <dtaht> I have to play with this 6lopan stuff.
[08:53:20] <dtaht> and this guy should go to battlemesh immediately after leaving the ietf today. battlemesh.org is running a test simultaneously with ietf
[08:53:22] <mcr> Dave: that's why I hedged my answer :-)
[08:54:21] <mcr> http://battlemesh.org/ has no IPv6 :-)
[08:54:23] sftcd leaves the room
[08:54:24] sftcd1 joins the room
[08:54:37] <dtaht> mcr: ULAs they have
[08:54:52] <dtaht> portable address space they want
[08:55:49] mcr is now known as mcharlesr
[08:55:49] mcharlesr is now known as mcr
[08:57:23] <SM> AS112
[08:58:15] <dtaht> yes, hop limits make sense
[08:59:10] <mcr> Dave Thaler.
[08:59:54] <dtaht> ... so long as they are wider than the network diameter.
[08:59:58] dtaht sits corrected
[09:00:50] mcr is now known as mcharlesr
[09:00:50] mcharlesr is now known as mcr
[09:01:52] <Andrew McGregor> +1 to Stuart
[09:02:14] <mcr> Stuart speaking.
[09:02:33] <Dave Thaler> in any network with actual multicast routing protocols, using a small hop limit is harmful. That's why administratively scope multicast address ranges were defined to solve.
[09:03:53] <dtaht> by 'trickle' is he referring to a specific product or just the concept?
[09:04:06] <wmtownsley> http://tools.ietf.org/html/rfc6206
[09:04:16] <dtaht> thx
[09:04:20] mcr channels sesame street and sings "Who are the printers in my neighourhood?"
[09:04:48] <SM> It depends on where the border is
[09:04:49] <mcr> was Tom Herbst.
[09:04:53] <Andrew McGregor> Nothing prevents you from running a mdns to regular dns proxy for the bits of network that have no multicast
[09:05:11] <Andrew McGregor> Or mdns multicast-to-unicast proxy
[09:05:13] markushx joins the room
[09:05:22] <dtaht> lots prevent you from a mdns proxy on a network with potentially multiple mdns proxys and multicast
[09:05:30] dtaht seriously melted down a network the other day
[09:05:51] mcr is now known as mcharlesr
[09:05:52] <mcharlesr> dthat, so multicast in some networks connected together by a proxied LLN?
[09:06:03] <mcharlesr> er, by two proxied LLN.
[09:06:15] <mcharlesr> trickle saves you a bit in the proxy.
[09:06:24] dtaht is admiring trickle
[09:06:38] <mcharlesr> Jeff Mulling (?)
[09:06:59] <dtaht> I melted down aformentioned network with avahi-daemon proxying mdns between 3 nets in the lab
[09:07:08] <Ralph Droms> Mulligan
[09:07:28] Peter Koch leaves the room: Computer went to sleep
[09:07:33] <mcharlesr> Jari Arkko.
[09:09:24] <mcharlesr> next talk: Homenet Security
[09:10:01] <mcharlesr> does guy in lower right have "RTFM" on his t-shirt?
[09:10:18] <Andrew McGregor> RTFM indeed
[09:10:23] jweil leaves the room
[09:10:52] mcharlesr is now known as mcr
[09:12:08] <mcr> slide 1. Make a list of knobs.
[09:13:08] <mcr> slide 2. Apply settings (policies) to the knobs
[09:14:51] <mcr> slide: Homenet's Goals for Policies
[09:15:25] <mcr> switches back to slide 1 and diagram.
[09:15:25] jweil joins the room
[09:15:30] <mcr> diagram.
[09:15:35] <mcr> ponting to printer in HOME network.
[09:15:53] mcr is now known as mcharlesr
[09:15:53] mcharlesr is now known as mcr
[09:16:52] <sftcd1> did he just say that a MITM SSL box was within architectural scope?
[09:16:54] <mcr> Ralph Droms going to mic.
[09:17:14] <mcr> sftdc1: he said that it was not in the scope of homenet, but that someone else could define advance security.
[09:17:25] <sftcd1> IMO MITM != advanced
[09:17:27] Wassim M. Haddad leaves the room
[09:17:58] <mcr> well, that would be a discussion about what is advanced security, and therefore is out of scope.
[09:18:17] <sftcd1> out of scope for that is good
[09:18:34] <mcr> "door is open" to do other things in the future.
[09:19:53] Alex Petrescu leaves the room
[09:20:22] <mcr> Dimitri.
[09:20:28] <mcr> slide: Homenet goals.
[09:20:53] mcr is now known as mcharlesr
[09:20:53] mcharlesr is now known as mcr
[09:21:11] Peter Koch joins the room
[09:21:54] <mcr> slide: diagram. points to main router in home.
[09:22:35] Atarashi Yoshifumi leaves the room
[09:23:47] <mcr> Tim Chown.
[09:24:32] Ole Troan leaves the room
[09:24:34] jpc leaves the room
[09:24:39] <mcr> (didn't catch his name)
[09:24:57] jpc joins the room
[09:25:06] <mcr> it was Toerless Eckert.
[09:25:53] mcr is now known as mcharlesr
[09:25:53] mcharlesr is now known as mcr
[09:26:24] <mcr> I think is Eric.
[09:26:46] <mcr> Jari Arkko at front.
[09:26:53] jweil leaves the room
[09:27:43] <mcr> Jari was running code while we were getting rough consensus! that's called multithreading!
[09:28:15] <dtaht> um, er, I've had a homenet like router running for over a year...
[09:28:31] <mcr> with ospf?
[09:28:33] <dtaht> and have had ospf running with quagga-re
[09:28:39] <mcr> I think you did quagga.... right?
[09:28:46] <dtaht> but rapidly went back to babel for the mesh routes
[09:28:59] <dtaht> quagga-re and now quagga do babel now
[09:29:11] <dtaht> so I - or those of you doing this -
[09:29:14] <dtaht> can easily try both now.
[09:29:24] <wmtownsley> So, when Jari finishes, if dtaht's implementation and his actually interoperate, we'll have two independent implementations. Yay.
[09:29:26] <dtaht> I AM NOT ready to ship quagga by default in cerowrt yet....
[09:29:46] <wmtownsley> s/independent/independent interoperable
[09:29:54] <dtaht> but quagga-re is packaged up into openwrt...
[09:30:01] <dtaht> (in the ceropackages repo)
[09:30:03] <dtaht> run amuck.
[09:30:08] <mcr> slide: Detailed Comments 1
[09:30:54] mcr is now known as mcharlesr
[09:30:54] mcharlesr is now known as mcr
[09:31:09] Chris Griffiths leaves the room
[09:31:24] <dtaht> or time.
[09:31:32] <mcr> slide: detailed comments 2
[09:31:45] <mcr> slide: detailed comments 3
[09:31:56] <mcr> slide: Protocol Values for testing
[09:31:58] <dtaht> https://github.com/Quagga-RE/quagga-RE
[09:32:00] Suzanne leaves the room
[09:32:25] Peter Koch leaves the room: Computer went to sleep
[09:32:39] woolf joins the room
[09:32:53] sftcd1 leaves the room
[09:33:04] Guillaume Habault leaves the room
[09:33:14] satoru.matsushima leaves the room
[09:33:21] gigix73 leaves the room
[09:33:24] teco.boot leaves the room
[09:33:28] Andrew McGregor leaves the room
[09:33:28] kawashimam leaves the room
[09:33:37] mcr leaves the room
[09:33:37] wmtownsley leaves the room
[09:33:39] iljitsch leaves the room
[09:33:40] bernie (ucom.ch) leaves the room
[09:33:43] SM leaves the room
[09:33:53] ogud leaves the room
[09:33:53] Ralph Droms leaves the room
[09:33:53] tsuichi leaves the room
[09:34:24] Thomas Heide Clausen leaves the room
[09:34:53] tsavo_work@jabber.org/Meebo leaves the room
[09:34:53] cabo leaves the room
[09:41:17] woolf leaves the room
[09:43:01] Daniel Gavelle [nxp] leaves the room
[09:43:18] cabo joins the room
[09:43:18] Klensin leaves the room
[09:48:40] wmtownsley joins the room
[09:49:25] markushx leaves the room
[09:50:53] Dave Thaler leaves the room
[09:50:53] monden leaves the room
[09:52:15] Daniel Gavelle [nxp] joins the room
[09:56:22] sftcd joins the room
[09:56:34] sftcd leaves the room
[09:56:34] jpc leaves the room
[09:57:15] Daniel Gavelle [nxp] leaves the room
[09:57:15] wmtownsley leaves the room
[09:57:50] jweil joins the room
[09:59:23] jweil leaves the room
[10:03:58] woolf joins the room
[10:04:59] woolf leaves the room
[10:05:21] bernie (ucom.ch) joins the room
[10:09:11] jweil joins the room
[10:09:20] bernie (ucom.ch) leaves the room
[10:09:23] jweil leaves the room
[10:11:47] bernie (ucom.ch) joins the room
[10:18:07] tsuichi joins the room
[10:23:09] Peter Koch joins the room
[10:24:05] Guillaume Habault joins the room
[10:27:16] Alex Petrescu joins the room
[10:29:56] bernie (ucom.ch) leaves the room
[10:33:43] Peter Koch leaves the room: Computer went to sleep
[10:34:28] bernie (ucom.ch) joins the room
[10:38:59] bernie (ucom.ch) leaves the room
[10:45:35] Alex Petrescu leaves the room
[10:45:35] Alex Petrescu joins the room
[10:47:23] Daniel Gavelle [nxp] joins the room
[10:47:50] Guillaume Habault leaves the room
[10:53:25] Daniel Gavelle [nxp] leaves the room
[10:55:05] cabo leaves the room
[10:55:40] Daniel Gavelle [nxp] joins the room
[10:56:20] Daniel Gavelle [nxp] leaves the room
[10:58:32] tsuichi leaves the room
[10:59:12] Ralph Droms joins the room
[10:59:57] mcr joins the room
[11:00:22] cabo joins the room
[11:01:35] Ralph Droms leaves the room
[11:03:34] mcr leaves the room
[11:04:35] Alex Petrescu leaves the room
[11:05:08] bernie (ucom.ch) joins the room
[11:05:08] bernie (ucom.ch) leaves the room
[11:05:18] bernie (ucom.ch) joins the room
[11:06:46] Daniel Gavelle [nxp] joins the room
[11:07:22] Daniel Gavelle [nxp] leaves the room
[11:07:30] wmtownsley joins the room
[11:10:41] Chris Griffiths joins the room
[11:13:08] Randall Gellens joins the room
[11:14:38] monden joins the room
[11:19:00] Chris Griffiths leaves the room
[11:24:05] monden leaves the room
[11:24:22] monden joins the room
[11:32:51] monden leaves the room
[11:32:56] monden joins the room
[11:33:04] monden leaves the room
[11:54:20] bernie (ucom.ch) leaves the room
[12:01:58] Alex Petrescu joins the room
[12:15:36] Alex Petrescu leaves the room
[12:16:08] Alex Petrescu joins the room
[12:17:47] bernie (ucom.ch) joins the room
[12:17:48] bernie (ucom.ch) leaves the room
[12:17:58] bernie (ucom.ch) joins the room
[12:21:18] wmtownsley leaves the room
[12:21:22] wmtownsley joins the room
[12:28:53] bernie (ucom.ch) leaves the room
[12:30:07] Alex Petrescu leaves the room
[12:42:04] Alex Petrescu joins the room
[12:43:34] Andrew McGregor joins the room
[12:50:37] Alex Petrescu leaves the room
[12:53:31] Randall Gellens leaves the room
[13:03:07] cabo leaves the room
[13:10:01] wmtownsley leaves the room
[13:13:12] cabo joins the room
[13:18:08] cabo leaves the room
[13:18:37] cabo joins the room
[13:24:20] wmtownsley joins the room
[13:51:14] Andrew McGregor leaves the room
[14:10:38] cabo leaves the room
[14:34:16] wmtownsley leaves the room
[14:53:07] cabo joins the room
[15:18:16] wmtownsley joins the room
[15:29:01] Andrew McGregor joins the room
[15:34:38] wmtownsley leaves the room
[16:05:16] ogud joins the room
[16:30:10] ogud leaves the room
[16:41:47] Peter Koch joins the room
[17:19:14] dtaht leaves the room
[17:22:15] dtaht joins the room
[17:31:46] Peter Koch leaves the room: Computer went to sleep
[17:34:11] cabo leaves the room
[17:35:58] Andrew McGregor leaves the room
[17:39:24] cabo joins the room
[18:08:41] cabo leaves the room
[20:50:49] Andrew McGregor joins the room
[20:54:51] Andrew McGregor leaves the room: Replaced by new connection
[20:54:52] Andrew McGregor joins the room
[21:11:01] Andrew McGregor leaves the room: Replaced by new connection
[21:11:05] Andrew McGregor joins the room
[21:24:09] Andrew McGregor leaves the room: Replaced by new connection
[21:24:11] Andrew McGregor joins the room
[21:27:43] Andrew McGregor leaves the room: Replaced by new connection
[21:27:44] Andrew McGregor joins the room
[21:43:18] Andrew McGregor leaves the room: Replaced by new connection
[21:43:18] Andrew McGregor joins the room
[21:48:17] Andrew McGregor leaves the room: Replaced by new connection
[21:48:27] Andrew McGregor joins the room
[21:54:14] Andrew McGregor leaves the room
[21:54:41] Andrew McGregor joins the room
[22:02:08] Andrew McGregor leaves the room
[22:31:12] wmtownsley joins the room
[22:41:10] wmtownsley leaves the room
[23:34:51] Thomas Heide Clausen joins the room
[23:36:47] Thomas Heide Clausen leaves the room
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!