[02:12:16] --- mrichardson has left
[09:06:51] --- mrichardson has joined
[09:06:57] --- mrichardson has left
[09:07:07] --- mrichardson has joined
[10:14:00] --- LOGGING STARTED
[10:16:47] --- mrichardson has joined
[10:19:28] --- mrichardson has left: Logged out
[10:19:36] --- mrichardson has joined
[12:59:30] --- LOGGING STARTED
[13:03:30] --- LOGGING STARTED
[15:36:42] --- mrichardson has joined
[18:31:51] --- iljitsch has joined
[18:39:21] --- juha.wiljakka has joined
[18:40:33] --- juha.wiljakka has left
[18:40:34] <iljitsch> anything happening? I hear chatter in the distant background on the audio stream...
[18:45:33] --- nm has joined
[18:45:54] --- nov has joined
[18:46:38] --- scottleibrand has joined
[18:46:51] <scottleibrand> geoff is talking now, can you hear him?
[18:46:58] <iljitsch> yes this is ok
[18:47:02] <scottleibrand> cool
[18:48:12] <iljitsch> please use the mike...
[18:48:43] <scottleibrand> that was just a comment on the use of "ietf 66" instead of 65
[18:48:55] --- brabson has joined
[18:49:03] <scottleibrand> and similarly unimportant stuff
[18:50:02] <iljitsch> Ah but that's just like in routing: how do I know I don't need to know if I don't know in the first place?
[18:50:17] --- scottleibrand has left
[18:50:30] --- mrw has joined
[18:52:16] --- momose has joined
[18:54:52] --- geir_egeland has joined
[18:55:55] --- scottleibrand has joined
[18:56:23] <scottleibrand> iljitsch: ya :)
[18:56:38] --- dmm has joined
[18:56:50] --- dlewis has joined
[18:57:09] <scottleibrand> this (physical) room is getting full; I'm impressed
[18:57:18] <iljitsch> how big is it?
[18:58:34] --- ggm has joined
[19:01:23] --- yushun has joined
[19:01:52] --- yushun has left
[19:02:43] <iljitsch> ugh
[19:02:51] <iljitsch> forking is evil
[19:03:26] --- momose has left: Replaced by new connection
[19:04:25] --- nm has left: Replaced by new connection
[19:04:47] --- nm has joined
[19:05:47] --- Tsubasa has joined
[19:06:21] --- nm has left: Replaced by new connection
[19:06:25] --- Tsubasa has left
[19:06:27] --- nm has joined
[19:06:27] --- nm has left
[19:06:50] <iljitsch> i haven't yet...
[19:07:11] <iljitsch> use bittorrent algorithm...
[19:07:41] --- momose has joined
[19:09:02] --- geir_egeland has left
[19:09:02] --- brabson has left
[19:09:27] <iljitsch> question: what about the issues that came up in NANOG? Such as TE which Geoff still referred to as an extension earlier.
[19:09:40] <dmm> I have a talk on that for today
[19:09:55] <iljitsch> dmm = ?
[19:10:06] <dmm> http://www.1-4-5.net/~dmm/IETF/IETF65/shim6
[19:10:07] --- geir_egeland has joined
[19:10:31] <iljitsch> I mean: who are you? :-)
[19:10:35] <dmm> I didn't, however, hear what Geoff said. I got here a bit late
[19:11:06] <iljitsch> nothing in particular, he just referred to a TE draft as an extension
[19:11:32] --- jc_lee has joined
[19:11:41] <dmm> David Meyer
[19:11:46] <iljitsch> ah ok
[19:12:29] <iljitsch> haven't we beaten the transport/network layer thing to death in multi6 already???
[19:12:35] --- Thierry has joined
[19:13:48] <iljitsch> I actually met people who expressed interest in implementing shim6 not too long ago...
[19:16:03] --- OleTroan has joined
[19:16:30] --- scottleibrand has left: Replaced by new connection
[19:16:34] --- scottleibrand has joined
[19:16:38] <iljitsch> Would we have fewer backward compatibility issues if we do experimental now and change some stuff when we go to standards track later rather than go for standards track immediately?
[19:17:03] <iljitsch> people are not saying their names when speaking, btw
[19:17:31] <iljitsch> THIS IS STUPID BECAUSE IT REQUIRES IMPLEMENTING DRAFTS THAT DON'T STICK AROUND!!!!!!
[19:18:00] <scottleibrand> wow, our wi-fi really sucked there for awhile
[19:18:24] <scottleibrand> i'll mention names
[19:18:27] <dmm> for a while?
[19:18:33] <scottleibrand> nm, he just did
[19:18:57] <scottleibrand> dmm: for quite awhile. :)
[19:19:05] <dmm> :-)
[19:19:17] <dlewis> Reply from 128.107.253.87: bytes=32 time=1068ms TTL=242 Reply from 128.107.253.87: bytes=32 time=294ms TTL=242 Request timed out. Request timed out. Request timed out. Request timed out. Request timed out. Reply from 128.107.253.87: bytes=32 time=648ms TTL=242 Reply from 128.107.253.87: bytes=32 time=179ms TTL=242
[19:19:20] <dlewis> ugh
[19:19:25] <scottleibrand> ya, I can handle latency
[19:19:32] <scottleibrand> it's the 99% packetloss that really sucks. :)
[19:19:32] <iljitsch> 802.11g yet?
[19:20:17] <iljitsch> New Apples also have A. :-)
[19:20:32] <scottleibrand> ya, i'm wishing I owned an A radio
[19:20:49] <scottleibrand> but didn't figure I could expense that as an IETF expense. :)
[19:23:54] --- mrichardson has left
[19:25:42] <iljitsch> so who is going to write this?
[19:26:01] <scottleibrand> sounds like it might already be started. marcello's looking
[19:26:42] <iljitsch> there may not even be anyone who is actually familiar with the details of shim6 AND mip6
[19:26:50] --- brabson has joined
[19:26:54] <scottleibrand> heh
[19:27:13] <dmm> a is way better, if you happen to have an a radio
[19:27:28] --- sivaram7 has joined
[19:27:31] <scottleibrand> ya, the NOC said it would be, but unfortunately I don't
[19:27:46] <iljitsch> g is good too if you can give it some room rather than have it pounded on by b.
[19:28:21] <scottleibrand> i wonder what fraction of clients are both b/g compatible nowadays
[19:28:30] <dmm> linux
[19:28:55] <scottleibrand> eh? i mean what fraction of radios/cards...
[19:29:09] <dmm> oh, don't know
[19:29:15] <iljitsch> many... but don't forget the hideous backward compatibiity that kicks in when there's even a single b node on a b/g network. you're better off running separate base stations.
[19:29:50] <scottleibrand> ah, then not much point in this environment, unless they want to do g-only and b-only stations
[19:29:51] <iljitsch> hideous backwards compatibility is something I'm very worried about for shim6 if we go too fast.
[19:30:02] --- nm has joined
[19:30:46] <iljitsch> the trouble is there aren't enough channels to run separate b and g networks concurrently...
[19:31:19] <scottleibrand> ya, unless you're willing to go with partially overlapping
[19:31:39] <scottleibrand> which they did at the L.A. nanog/arin and it worked pretty well. i don't know if they seperated b and g tho
[19:31:40] <iljitsch> which you should always do, that's what it's designed for.
[19:32:20] <scottleibrand> ya, I guess the question is how much overlap between adjacent nodes is ok to minimize closer spectrum overlap with more distant nodes
[19:33:16] <scottleibrand> i wonder if mesh networking could improve things by letting everyone lower their transmit power to only talk to the X nearest nodes
[19:33:18] <iljitsch> there is a lot of research going on in this and related areas...
[19:33:26] --- nm has left: Replaced by new connection
[19:33:26] --- nm has joined
[19:33:26] --- nm has left
[19:33:42] <scottleibrand> ya I'm looking forward to manet
[19:33:47] <iljitsch> ...especially that one... go to infocomm06 or read the proceedings if you're interested.
[19:33:59] --- nm has joined
[19:34:44] --- momose has left: Replaced by new connection
[19:35:13] <scottleibrand> how does infocomm compare/relate to supercomm (now globalcomm)?
[19:35:54] <iljitsch> don't know, I got drafted in infocomm tpc but other than that those aren't really my circles.
[19:36:18] <scottleibrand> when they had supercomm in atlanta I attended, and they had quite an interesting talk on mesh networking
[19:36:32] --- OleTroan has left
[19:37:05] <iljitsch> lots of interesting stuff, but what I got away with is the fact that backward compatibility is a major pain in the rear side.
[19:37:20] --- sivaram7 has left
[19:37:34] <scottleibrand> you're referring to backwards compatibility for mesh networking, or something else?
[19:37:43] <iljitsch> (wrt to ieee 802.11 improvements)
[19:37:50] <scottleibrand> ya
[19:41:06] --- brabson has left
[19:42:27] --- dmm has left
[19:43:51] <scottleibrand> marcelo just posted the shim-mip interaction draft url to the wg list
[19:44:00] <scottleibrand> http://www.watersprings.org/pub/id/draft-bagnulo-shim6-mip-00.txt
[19:44:21] <iljitsch> while he's talking? cool.
[19:44:22] --- brabson has joined
[19:44:55] <scottleibrand> probably he posted it before he started
[19:44:59] <scottleibrand> i think the list is slow
[19:45:20] <scottleibrand> ya the date is 18:34 -0600
[19:46:40] <iljitsch> don't forget that there's no multihoming privacy today either.
[19:47:19] --- FDupont has joined
[19:48:20] <scottleibrand> iljitsch: let us know if you want us to repeat anything to the mic
[19:48:35] <iljitsch> hmmmm ok
[19:48:37] <scottleibrand> i don't think anyone is officially transcribing, but I'd volunteer on a case-by-case
[19:49:27] --- rbless has joined
[19:51:00] --- gih900 has joined
[19:51:49] --- dumdidum has joined
[19:52:04] <gih900> if you could transcribe that would be appreciated - noone wanted to do it when we started off the wg meeting :-(
[19:52:36] --- dmm has joined
[19:52:39] <iljitsch> is transcribing to jabber useful in these modern audiostreaming times??
[19:52:44] <ggm> YES
[19:52:53] <ggm> I am in rtgarea. this feed lets me know whats happening
[19:53:04] <iljitsch> I rather use jabber as an out of band channel.
[19:53:17] <iljitsch> nothing. :-)
[19:53:51] <ggm> oh, even that is useful.
[19:53:57] <scottleibrand> I was talking about unidirectional jabber->mic communication
[19:54:16] --- geir_egeland has left
[19:54:16] <scottleibrand> are you asking for audio->jabber transcription also, or just question/comment passalong?
[19:54:24] --- nm has left: Replaced by new connection
[19:54:29] <ggm> I think unidirectional jabber/mike is done in UDLR
[19:54:40] <iljitsch> reachability protocol doesn't allow unidirectional communication. :-) (ok I'll stop the silliness now)
[19:55:09] <gih900> I think I was asking just for notes of highlights of themicrophone conversation rather than an attempt to ttranscribe the commentary against the slidepacks
[19:59:35] --- scottleibrand has left: Replaced by new connection
[19:59:43] --- bert has joined
[20:02:20] --- nm has joined
[20:02:40] <iljitsch> next on dave meyer (myer?) his slides are in the proceedings.
[20:04:11] <iljitsch> operator issues with shim6 as heard on nanog and appricott
[20:04:42] --- scottleibrand has joined
[20:05:15] <scottleibrand> wee I love wi-fi. sounds like iljitsch is doing some transcription now?
[20:05:38] <iljitsch> not transcribing, just noting what's happening from time to time
[20:05:46] <scottleibrand> k
[20:05:54] <scottleibrand> i can help if you'd like
[20:06:23] <iljitsch> slides are online, people can look at those... maybe do some mike comments?
[20:06:44] <scottleibrand> k
[20:07:02] <scottleibrand> and anything I notice him saying that's not on the slides
[20:07:33] <iljitsch> multi6 and shim6 are also site based
[20:08:59] <iljitsch> I have a question for Dave when he's finished: are they putting on this show at the next Istanbul or Amsterdam RIPE meeting also?
[20:09:36] --- miyahiro has joined
[20:10:57] <gih900> yes - I believe that Dave is heading to Istanbul - but do you want me to ask?
[20:11:09] <iljitsch> if it doesn't come up, yes
[20:11:31] <gih900> it came up - yes
[20:12:01] <iljitsch> wow lag is pretty bad, I see your comment 20 seconds before I hear the audio
[20:12:22] <scottleibrand> ok. I'll ask about RIPE.
[20:12:29] <gih900> no - I looked at the slide before he spoke!
[20:12:35] <iljitsch> ah
[20:12:49] <iljitsch> amsterdam or istanbul?
[20:12:56] <gih900> istanbul
[20:13:05] <scottleibrand> just asked, and he said yes to ripe amsterdam
[20:13:25] <iljitsch> ok looks like I have to be at both istanbul and amsterdam then...
[20:13:32] <scottleibrand> heh
[20:13:35] <gih900> RIPE 52 (late April) is in Instanbul - Dave will be going to Istanbul whether he knows it now or not! :-)
[20:13:48] --- nm has left: Replaced by new connection
[20:13:49] --- nm has joined
[20:13:49] --- nm has left
[20:14:21] <scottleibrand> heh, that's probably what he meant then
[20:14:33] <iljitsch> don't forget that nanog is the same people who had a conniption over unique site locals which by definition don't involve isps.
[20:14:34] <scottleibrand> i said amsterdam based on my quick read of your comments, he never said it.
[20:15:02] <scottleibrand> iljitsch: nanog (the conference) != nanog (the mailing list)
[20:15:15] <scottleibrand> IMO the former is much more reasonable.
[20:15:28] <iljitsch> I did notice some overlap though...
[20:15:49] --- nm has joined
[20:16:23] <scottleibrand> ya, that's to be expected. the level of hysteria is much higher on the list though, to the point where I permantently unsubscribed
[20:17:01] <iljitsch> nanog volume is constant, snr varies depending on whether something of relevance is happening in the world
[20:17:17] --- dlewis has left
[20:18:21] --- miyahiro has left
[20:19:08] --- scottleibrand has left: Replaced by new connection
[20:19:53] --- dmm has left
[20:20:12] <iljitsch> my laptop has 6 interfaces. :-)
[20:22:14] --- washad has joined
[20:24:01] --- scottleibrand has joined
[20:24:36] <scottleibrand> heh nice. how'd you manage to get 6 int's on a laptop?
[20:24:59] <scottleibrand> mine only has one pcmcia card, two usb ports, and two built-in (wi-fi and wired)
[20:25:29] <iljitsch> erik nordmark is now talking about extensions to the base shim6 mechanisms which would allow a full loc/id split (where the id isn't one of the locs, now discussing lookup mechanisms) and I think he has stuff about traffic engineering too.
[20:26:05] <iljitsch> hm forgot about pcmcia. mine has usb, firewire, wifi, ether, bluetooth and a modem
[20:26:26] <iljitsch> and I run Zebra on it. :-)
[20:26:37] <scottleibrand> heh nice
[20:26:53] <scottleibrand> i do that with some desktop boxes, but my laptop is too important to mess around with its routing too much. :)
[20:27:03] <iljitsch> yes apple stuff is too expensive but you do get some good stuff
[20:27:11] <scottleibrand> the extent of my messing around is pointing my default gateway at a zebra box
[20:28:04] <scottleibrand> ugh that SRV stuff looks ugly
[20:28:43] <iljitsch> yes but do we want to invent our own DNS RRs? DNS guys may run us out of town even with out that
[20:29:08] <scottleibrand> i'm wondering what benefits a full loc/id split gives us over the partial split in shim6
[20:29:54] <iljitsch> stable addressing. people REALLY want this and this way it doesn't make us all buy bigger TCAMs
[20:30:36] <scottleibrand> does stable ID addressing help if your locator addressing still requires renumbering and reconfiguration?
[20:32:05] <iljitsch> well i suppose it's not as easy as PI addressing but it's still a lot better than "real" renuumbering. it's not only the network stuff, though: apps people don't like the fact that in ipv6 you can have multiple addresses that can come and go at any time and you have to try them all
[20:32:30] <iljitsch> traffic engineering!
[20:32:46] <scottleibrand> ok. sounds like stuff that can be addressed without a full id/loc split to me, but that conversation can wait.,
[20:32:46] <iljitsch> we already have priority and weight in shim6 signalling as soon as that's ddone
[20:33:39] <scottleibrand> ya, the problem is getting those values properly autoconfigured
[20:35:51] <scottleibrand> i have a question about something I've been thinking about recently: would it be possible to configure a shim6-capable host to send packets over both locator pairs at once in order to fully utilize two links for a single session (aggregating bandwidth)?
[20:36:04] <scottleibrand> the sending host would have to keep two tcp congestion windows, of course...
[20:36:45] <scottleibrand> wee source rewriting. this is also interesting stuff.
[20:37:02] <iljitsch> then tcp would have to know about all of this. I think that's a logical next step when shim6 is in place. It would be very cool because you get to utilize the full combined bandwidth of all your links.
[20:37:17] <scottleibrand> ya, that's what I was thinking.
[20:37:17] --- dumdidum has left: Replaced by new connection
[20:37:29] --- dumdidum has joined
[20:37:32] <scottleibrand> agreed it would have to heavily involve tcp
[20:38:08] <iljitsch> tcp and ip are implemented together generally anyway, i gather, so in practice that would happen anyyway.
[20:39:10] <scottleibrand> ya that's what I was thinking. i noticed they're very separate in IETF tho.
[20:39:27] --- nm has left: Replaced by new connection
[20:39:28] --- nm has joined
[20:39:28] --- nm has left
[20:39:59] <scottleibrand> i really like the idea of the shim6 hosts learning router preferences based on how the source locators are rewritten.
[20:41:48] <iljitsch> ugh, isn't the point of encryption that stuff in the middle CANNOT see inside your packet???
[20:42:01] <scottleibrand> lots of back-and-forth as to whether source locator rewriting breaks ipsec
[20:42:31] <iljitsch> you could do ipsec on top of shim or underneath. doesn't matter as long as both ends agree
[20:42:49] <scottleibrand> ya, i think the related question is whether shim6 is "inside" the packet or "outside" the encryption
[20:42:49] <iljitsch> if you do it underneath you had better not use AH and rewrite.
[20:43:03] <iljitsch> ESP doesn't care
[20:43:29] * scottleibrand is glad there are other people who are willing to take the time to fully understand security
[20:43:52] <scottleibrand> i get lost easily when discussions get into this much detail
[20:43:52] <iljitsch> yeah "fully"...
[20:43:55] --- nm has joined
[20:44:29] --- rbless has left
[20:45:48] * scottleibrand wonders if he could patent a square wheel
[20:46:10] * iljitsch wishes he had a patent on spam
[20:47:22] <iljitsch> well if we can have routers rewrite, why would NATs be a problem?
[20:47:24] --- scottleibrand has left: Replaced by new connection
[20:47:32] --- scottleibrand has joined
[20:47:47] <scottleibrand> non-shim6-aware NATs
[20:48:10] <iljitsch> anonymous at the mike: hey if you do all of this it starts to look a lot like HIP except that the crypto bit is bigger so lookup is harder
[20:48:22] <gih900> Dave Thaler at the mic
[20:48:37] --- scottleibrand has left: Replaced by new connection
[20:48:42] --- scottleibrand has joined
[20:49:21] <iljitsch> hm but router rewriting and nat rewriting are probably similar enough that we can accommodate both without too much trouble if we care, and then we can do shim4.
[20:50:09] <iljitsch> but security is a problem with v4 because there's no hba/cga. so maybe support pki too?
[20:50:22] --- bert has left
[20:50:38] --- washad has left: Computer went to sleep
[20:50:40] <scottleibrand> non-shim6-aware NATs would be a problem, no?
[20:51:53] <iljitsch> hm if security and signalling work we really don't care much about the addresses... probably need to slap on a udp header in v4 to enable nat, though.
[20:51:57] <scottleibrand> at the very least, I could see destination locator rewrite as a possible problem...
[20:52:38] <scottleibrand> ya, the protocol 40 or whatever shim6 control packets become wouldn't get through most likely
[20:52:48] --- mrw has left
[20:52:52] --- brabson has left
[20:54:05] --- nov has left
[20:55:00] --- jc_lee has left
[20:55:14] <iljitsch> and we're done!
[20:55:38] --- gih900 has left
[20:59:38] --- iljitsch has left
[21:01:08] --- ggm has left
[21:01:08] --- scottleibrand has left: Replaced by new connection
[21:04:51] --- mrichardson has joined
[21:12:35] --- nm has left
[21:12:51] --- dumdidum has left
[21:13:33] --- FDupont has left
[21:18:29] --- Thierry has left: Replaced by new connection
[21:18:30] --- Thierry has joined
[21:40:46] --- mrichardson has left: Logged out
[22:23:30] --- Thierry has left
[23:33:09] --- nm has joined
[23:33:23] --- nm has left
[23:34:37] --- nm has joined
[23:36:50] --- nm has left