[09:17:09] --- keithmoore has joined
[09:45:39] --- wouter has joined
[09:45:53] <wouter> Hi Keith
[09:46:05] <keithmoore> hi
[09:54:05] --- bew has joined
[09:54:25] --- jeffa has joined
[09:55:05] --- Yuzhi Ma has joined
[09:56:45] --- sunakichiwide has joined
[10:01:39] --- lars has joined
[10:01:43] --- gurtov has joined
[10:01:59] --- eludom has joined
[10:02:01] <gurtov> gonzalo shows agenda
[10:02:08] --- ShoichiSakane has joined
[10:02:26] <gurtov> have a look at supplemental web page
[10:03:00] <gurtov> arch draft in Editor queue
[10:03:09] <gurtov> RFC4423 its number
[10:03:32] <gurtov> security area person will do analysis
[10:03:47] <gurtov> charley kapfman?
[10:04:25] <gurtov> discussing the status of all drafts
[10:04:56] <gurtov> mm draft under WGLC
[10:05:05] --- inet6num@jabber.org has joined
[10:05:21] <gurtov> other charter items are past LC already
[10:05:29] <gurtov> please comment mm
[10:06:01] <gurtov> showing changes from last draft versions - only minor changes
[10:06:29] <gurtov> rechartering topics: nat traversal and HIP applications
[10:06:48] --- dthaler has joined
[10:07:05] <gurtov> Miika Komu started NAT talk
[10:07:30] <dthaler> (note: spelling is Charlie Kaufman, up above)
[10:07:43] <gurtov> outline of the talk, motivation, packet formats, base exchange, firewall
[10:08:00] <dthaler> :q
[10:08:07] --- petri.jokela has joined
[10:08:19] --- andrewdmcgregor@jabber.psg.com has joined
[10:08:44] <gurtov> motivation: practical and implementable draft
[10:08:54] <andrewdmcgregor@jabber.psg.com> Do we have a notetaker?
[10:09:17] <jeffa> Pekka is taking notes, Andrei is jabber scribing
[10:09:17] <gurtov> pekka n takes notes
[10:09:21] <andrewdmcgregor@jabber.psg.com> thx
[10:10:12] <gurtov> nat detection is done using external protocols - no changes to HIP drafts
[10:10:33] <gurtov> implementations ongoing in NEC and HIIT
[10:10:51] <gurtov> HIP control channel header format
[10:11:12] <gurtov> HIP checksum set to zero
[10:11:14] --- dthaler has left
[10:12:03] <gurtov> ESP encapsulation from other groups
[10:12:21] <gurtov> picture of base exchange over UDP
[10:12:27] --- vivien.schmitt has joined
[10:13:17] <gurtov> about UDP ports: separate for control and data
[10:14:02] <lars> andrei at the mike
[10:14:45] <gurtov> AG: asked if UDP ports are available and could be reserved from iana
[10:15:15] <gurtov> the answer was yes
[10:15:36] <lars> andrei: i think there's text in there that says "upon publication, reserve two ports"
[10:16:17] <gurtov> ok, but publication is not required for high ports for reservation, we can do it now
[10:16:32] <gurtov> use scenarios for mobility
[10:17:03] <gurtov> multihoming more complicated than mobility
[10:17:37] <gurtov> not yet handled in draft
[10:17:44] <gurtov> erik nordman at mike
[10:18:09] <gurtov> if host has one interface but dst several?
[10:18:18] <gurtov> need two mapping in the nat
[10:18:40] <gurtov> MK continues on firewall configuration
[10:19:23] <gurtov> required firewall policies
[10:19:37] <gurtov> open issues in the tracker
[10:19:54] <gurtov> #1 use the same port numbers as IKE
[10:20:15] <gurtov> benefit: no extra firewall config
[10:20:30] <gurtov> Pekka Nikander at mike
[10:20:37] --- sarolaht has joined
[10:20:48] <gurtov> PN: two sides
[10:21:31] <gurtov> PN: is it desirable? define both scenarios in the draft
[10:21:58] <gurtov> PN: IKE uses reserved SPI to demultiplex packets
[10:22:13] <gurtov> MK: how to separate IKE and HIP packets
[10:22:33] <gurtov> PN: register another SPI<256 from iana
[10:22:51] <gurtov> PN: some people will have to change their firewall rules to disable HIP
[10:23:08] <gurtov> PN: suggest defining both ways in the draft
[10:23:27] <gurtov> andrew mcgregor at mike: require both for now
[10:23:48] <gurtov> david ward: sounds like a good idea
[10:24:01] <gurtov> MK: ok
[10:24:21] <gurtov> lars eggert: not sure if registered UDP ports can be reused
[10:24:32] --- jlaganie has joined
[10:24:53] <gurtov> PN: not IKE but separate demultiplex model need changes
[10:25:26] <gurtov> PN: earlier arguments if ESP can be reused for HIP
[10:25:53] <gurtov> PN: somebody argued that HIP can use packet format but not protocol number
[10:26:15] <gurtov> LE and AM at mikes
[10:26:58] --- alfredo.matos has joined
[10:27:01] <gurtov> ?: current IANA policies
[10:27:28] <gurtov> ?:operating through the NAT - is the a keepalive
[10:27:39] <gurtov> MK: yes
[10:27:57] <gurtov> AM: current iana policy requires publishing an rfc
[10:28:14] <gurtov> gonzalo: check the type
[10:28:25] <gurtov> ?: make survery of implementations
[10:28:42] <gurtov> MK: issue #3 server behind NAT
[10:29:02] <gurtov> need help of 3rd host
[10:29:34] <andrewdmcgregor@jabber.psg.com> rfc2406 does not specify what sort of RFC is required to allocate a reserved SPI; therefore any RFC should do. There are only two reserved values at present.
[10:30:01] <gurtov> for p2p friendly NAT, UDP hole punching is primary method
[10:30:33] <gurtov> gonzalo: check TURN reference
[10:30:33] <lars> meets later today
[10:30:39] <lars> behave, that is
[10:31:05] <gurtov> design alternative: ICE, maybe an overkill
[10:31:10] --- wouter has left: Replaced by new connection
[10:31:11] <gurtov> SIP oriented
[10:31:42] <gurtov> gonzalo: had this discussion on ICE for signalling. cannot do this
[10:32:15] <gurtov> open issue: server behind a nat as separate draft?
[10:32:44] <gurtov> in any case client and server should be 100% compatible
[10:32:53] <jlaganie> julien: having it in two drafts is overhead
[10:33:05] <gurtov> AM: both client and server are not possible
[10:33:38] <jlaganie> julien : could define rendezvous extensions in NAT traversal to allow for server behind NAT
[10:34:00] <gurtov> AM: details of NAT behavior differ depending on implementations
[10:34:08] <andrewdmcgregor@jabber.psg.com> To clarify: both ends behind nats is not *always* possible, sometimes you require a third party.
[10:34:25] <gurtov> gonzalo: need define the scope
[10:34:41] <gurtov> LE: rechartering items
[10:35:13] <gurtov> MK: open issue #4 - breaks rendezvous server
[10:35:43] --- ogud has joined
[10:35:54] <gurtov> NAT drops R1 from RVS
[10:36:23] <gurtov> AM: few bits for NAT type?
[10:37:00] <gurtov> MK: issue #5 - address types in LOCATOR
[10:37:17] <gurtov> several alternatives
[10:37:23] --- jasso1 has joined
[10:37:49] <gurtov> MK alternative 3 might be best
[10:38:02] <gurtov> PN: didn't understand it
[10:38:22] <gurtov> MK: before handover send update with addresses
[10:38:39] <gurtov> LE: depend on mobility implementation using outer addresses?
[10:38:48] <gurtov> AM above
[10:39:03] <gurtov> AM: need to experiment with various NATs
[10:39:03] <jeffa> here's that issue tracker: http://hip4inter.net/cgi-bin/roundup.cgi/hip-nat/index
[10:39:38] <gurtov> Erik Nordman: no semantic differences between private and public addresses
[10:39:48] <gurtov> MK: good point
[10:40:19] <gurtov> MK: issue 6 - inner address ipv4
[10:40:32] <gurtov> MK: not an issue anymore
[10:40:37] --- alban.hessler has joined
[10:40:40] --- jasso1 has left
[10:40:59] <gurtov> issue #7 - editorial notes
[10:41:24] <gurtov> issue #8 - mobility and data channel reactivation
[10:41:54] <gurtov> solution - use the same UDP port for control and data
[10:42:08] <gurtov> somebody opening a coke :)
[10:42:53] <gurtov> PN: will end up using one port number
[10:43:35] <gurtov> MK: vivien experimented with other approach
[10:43:56] <gurtov> MK: need to discuss on the list
[10:44:17] <gurtov> issue #9 - hairpin translation
[10:44:32] <gurtov> both hosts are behind the same nat
[10:45:24] <gurtov> PN: two years ago made recommendation so that always first send non-UDP packet
[10:45:37] <gurtov> PN: makes sence from architectural point of view
[10:46:06] <gurtov> MK: ok, good point
[10:46:14] <gurtov> MK: questions?
[10:46:42] <gurtov> Gonzalo: have a discussion on rechartering
[10:47:09] <gurtov> nat traversal, HIP applications
[10:47:27] <gurtov> documents relevant to nats
[10:47:40] <gurtov> 4 drafts
[10:48:01] <gurtov> the charter was let people experiment with HIP
[10:48:21] <gurtov> only something conrete, not research
[10:48:54] <gurtov> not developing API, but study the use of existing ones
[10:49:04] <gurtov> opening mike for comments
[10:49:47] <gurtov> AG: asked about native API
[10:50:32] <gurtov> AM: need implementation experience with UDP draft
[10:51:01] <gurtov> david: the document lays out all NAT issues
[10:51:50] <gurtov> AM: will discover new things during experimentation and feed to the draft
[10:52:24] <gurtov> gonazalo: have a look at behave WG
[10:52:54] <gurtov> ?: publishing API - revisit that if it's important
[10:53:12] <gurtov> david: were discussing that for six months
[10:53:57] <keithmoore> ietf _does_ do APIs. it doesn't standardize APIs as the basis for interoperation. interoperation is based on bits on the wire.
[10:54:01] <gurtov> jeff houston: if you need API to make it work - do it!!
[10:54:53] <gurtov> weng from ISI: support api
[10:55:01] <lars> yushun wang
[10:55:39] <keithmoore> (RAISE HAND)
[10:55:53] <gurtov> david made a pole
[10:56:10] <gurtov> API will be taken to the charter based on consensus
[10:56:23] <lars> what was the result on hip/nat? in terms of the new charter?
[10:56:32] <gurtov> MK: propose additional workin item - BOS support
[10:57:38] --- lars has left: Logged out
[10:57:39] <gurtov> MK: also consider advanced stuff on multiple SAs or is this too researchie?
[10:58:01] <gurtov> david: can be done in WG
[10:58:25] <gurtov> PN: basic mechanisms should not prevent advanced stuff
[10:59:39] <gurtov> PN: too earlier to fix alternatives e.g. in multihoming
[11:00:33] <gurtov> Lars: NAT would be taken to the charter
[11:01:00] <gurtov> MK: maybe too earlier but there was some work with shim6
[11:01:04] --- ogud has left
[11:01:11] <gurtov> MK: draft with shim6 interop?
[11:01:24] <gurtov> PN: david teller took it up on shim6
[11:01:38] <gurtov> PN: something will happen there, that's enough for us
[11:02:18] <gurtov> erik nordman: hip - shim6 interworking - header format for software portability
[11:02:59] <gurtov> geoff huston, shim6 co chair
[11:03:01] --- bew has left: Disconnected
[11:03:22] <gurtov> GH: document moved to LC based on discussions with MIPv6 only
[11:03:49] --- jan has joined
[11:03:58] <gurtov> GH: no serious issues, some basic alignment of protocols
[11:04:13] <gurtov> GH: but interopebility is not on shim6 charter
[11:04:27] <gurtov> GH: encourage taking it to HIP
[11:04:48] <keithmoore> (side comment) it seems like a shame that HIP gets less attention due to the fact that it's more widely applicable / versatile than other protocols that are trying to attack narrower problems...leads to piecemeal solutions.
[11:04:57] <gurtov> AM: story about shim6 mobility - a research story
[11:05:43] <gurtov> GH: shim6 locator agility not mobility
[11:06:09] <andrewdmcgregor@jabber.psg.com> I misspoke
[11:06:34] <gurtov> david: we're concluded
[11:06:38] --- keithmoore has left
[11:07:05] --- ShoichiSakane has left
[11:07:27] <gurtov> ? above = Mark Townsley, AD
[11:07:38] <jlaganie> Is the meeting done?
[11:07:42] --- Yuzhi Ma has left
[11:07:43] <gurtov> yes
[11:08:43] --- alban.hessler has left
[11:09:46] --- sunakichiwide has left
[11:10:12] --- vivien.schmitt has left
[11:10:25] --- inet6num@jabber.org has left
[11:11:24] --- gurtov has left
[11:11:30] --- jeffa has left
[11:13:01] --- jan has left: Logged out
[11:13:35] --- petri.jokela has left
[11:16:39] --- jlaganie has left
[11:18:07] --- sarolaht has left
[11:18:36] --- sunakichiwide has joined
[11:18:46] --- ogud has joined
[11:19:01] --- ogud has left
[11:19:57] --- sunakichiwide has left
[11:26:03] --- alfredo.matos has left
[11:27:19] --- andrewdmcgregor@jabber.psg.com has left
[11:45:18] --- bew has joined
[11:52:29] --- bew has left: Disconnected
[12:08:31] --- washad has joined
[12:10:51] --- LARS has joined
[12:11:30] --- washad has left
[12:12:02] --- LARS has left
[12:25:04] --- eludom has left