[10:09:22] --- nevil has joined
[10:10:08] --- nevil has left
[11:56:21] --- LOGGING STARTED
[13:28:02] --- gerhard has joined
[13:28:08] --- gerhard has left: Disconnected
[13:38:20] --- gerhard has joined
[13:56:56] --- nevil has joined
[14:03:27] --- oak has joined
[14:03:47] --- Simon Leinen has joined
[14:07:55] --- irino has joined
[14:08:25] <nevil> Juergen is reviewing the current IPFIX and PSAMP draft swtatus
[14:09:24] --- andrjohn has joined
[14:12:27] <nevil> Possible new charter items - presentations as per agenda (slide are on ietf meeting materials page)
[14:12:57] <nevil> consider whether they could/should be WG work items
[14:15:33] <nevil> Dan Romascanu: new work items - need to show there's a *need* for it, good to have *code*, need to have *3* people to work on it !
[14:15:48] --- Simon Leinen has left
[14:22:55] <nevil> Paul AItken presented changes (from WGLC) to testing draft [sorry, I was showing his slides]
[14:23:09] <nevil> Now Benoit is presenting the IPFIX MIB draft
[14:28:14] <nevil> Brian Trammell, SCTP Stream Restriction Change draft
[14:35:11] <gerhard> hum
[14:35:57] <nevil> Juergen asks "do people agree we need to change protocol?
[14:36:28] <nevil> Dave Plonka asks 'how can you say there's code that uses this?
[14:36:39] <gerhard> (my hum was related to the general necessity to change something)
[14:37:08] <nevil> BT: don't know wther there's code
[14:37:34] <nevil> Elisa Boschi: have had interop events, people asked "why is there this restriction?
[14:37:59] <nevil> -- points out need for change
[14:38:06] <gerhard> for the protocol: there has been an abundant discussion on this topic on the mailing list where a lot of arguments in favor of the change have been mentioned
[14:39:36] <nevil> Elisa says 'two implementations have already made changes in code
[14:40:00] <nevil> Benoit reminds us that we thought SCTP set reliability per- stream
[14:40:34] <nevil> Juergen: Are there any tech problems with the change?
[14:41:23] <nevil> Paul Aitken: only problem is that people may miss the change if they only read protocol dov
[14:41:45] <nevil> Agreed change is needed. Dan on how to make it ..
[14:42:05] <nevil> Proto doc is in RFC Editor Queue
[14:43:05] <nevil> Could pull it back, change it, do short WG last call, run it through IESG again. Est time, 2 months. This would be the best way
[14:45:24] <nevil> Testing and Implementation Guidelines will also need changes, but they're still in LC
[14:48:02] <nevil> Can protocol editors make changes? Benoit suggests working on those this week
[14:49:15] <nevil> Meanwhile we'll send proposal to make these changes (and only the SCTP use changes) to list for brief (1 week) comment, then will go ahead
[14:49:57] <nevil> Elisa says implementation changes can be done then also. Draft editors will meet later this week
[14:50:19] <nevil> Now Benoit on 'Export-per-SCTP-stream
[14:53:06] <nevil> - using single template/stream allows finding loss rate for that template
[14:54:58] <nevil> having new templates created on-the-fly by routers could need lots of streams
[15:00:03] <nevil> kobayashi: collector has to direct data to proper template always
[15:00:35] <nevil> benoit points out that collector may have to cache data records until template has arrived
[15:01:44] <nevil> Dan: let's discuss intended status
[15:02:54] <nevil> benoit would prefer stds track for this work item, but would that mean you *had* to do it that way
[15:03:20] <nevil> want to keep it as an option
[15:03:36] <gerhard> I'm not convinced that this should be a standard track document, rather informational.
[15:04:08] <nevil> Dan [contributor comment]: refining the protocol doesn't seem right
[15:04:16] <nevil> Dan: WG must decide
[15:04:42] <nevil> benoit wants this an an option in the protocol
[15:04:56] <nevil> Gerhard saya it would be better as an info doc
[15:05:27] <nevil> Elisa says it's similar to 'reducing redundancy'. Benoit agrees
[15:06:25] <nevil> Dave P asks "if you've found a protocol problem (can't always get loss rate accurately), why not fix the protocol?
[15:06:49] <nevil> Benoit: spec doesn't say MUST, only asks for an estimated loss rate
[15:07:17] <nevil> Dan suggests Experimental doc status - if it's useful it could change to Proposed Std
[15:08:40] <nevil> Humm - moderate support
[15:09:07] <nevil> Now Benoit on Config Data Model
[15:10:49] <nevil> Input from IPFIX and PSAMP MIB work
[15:11:53] <nevil> There is running code for this
[15:12:24] <nevil> Also,Cisco have done something similar for flexible NetFlow
[15:16:41] <nevil> Dave Harrington: if you want it to use XML, maybe you should use netconf ?
[15:17:03] <nevil> .. could end up re-writing it later to be compatible with netconf
[15:17:24] <nevil> .. netconf is proceding, still a way to go
[15:18:43] <nevil> Simon Leinen: encourage XML config model, but thinks Dave was saying "let the professionals make a protocol"
[15:18:54] <nevil> .. but we don't really want to wait too long
[15:19:45] <nevil> Dan: XML is OK, but go to BOF on data modelling later this week / next IETF
[15:20:03] <nevil> .. maybe IPFIX will be a first in XML config!
[15:20:27] <nevil> BOF Thursday. Dave has a BOF on "reading MIBs via XML"
[15:21:07] <nevil> .. write your requirements in English, then can move it to whatever XML format is needed later
[15:21:27] <nevil> Juergen: did this for IPFIX Info Model
[15:23:41] <nevil> Agreed, IPFIX should work on this. (clearly it's along-term effort though)
[15:23:51] <nevil> Now, BT on IPFIX File Format
[15:24:53] <nevil> Draft stable since Montreal, some changes at Prague
[15:26:34] <nevil> BT presents changes in draft
[15:30:41] <nevil> FIle Fmt adopted as WG item
[15:31:17] <nevil> Elisa on "Extended Types" i.e. need for info about Enterprise Info Elements
[15:33:50] --- m_ersue has joined
[15:34:04] --- m_ersue has left
[15:35:28] <nevil> benoit: only for Enterprise IEs?
[15:35:46] <nevil> Elisa: Yes in current draft, could be extended
[15:36:23] <nevil> Benoit:OK for Enterprise IEs, don't think so for Standrad IEs
[15:38:15] <nevil> Emile Stephan: doesn't see need to do this (?)
[15:39:52] <nevil> Benoit: Emile wants something like what David Moore proposed in 2005 - overload IP Address. That would be dangerous
[15:41:11] <nevil> Elisa: problem is not just related to file draft
[15:41:33] <nevil> Move this discussion to mailing list
[15:42:37] <nevil> Kobayashi on IPFIS Mediators
[15:50:33] <nevil> Benoit says there's a need fir this
[15:50:48] <nevil> Kobayashi: we have an implementatiom
[15:51:11] <nevil> Emile: We (France Telecom) need it too
[15:54:18] <nevil> Irino: performance improvement from element ordering
[15:55:51] <nevil> .. slide shows measured improvement
[15:57:07] <nevil> Paul Aitken: would like to see others repeat this test
[15:58:03] <nevil> Tanja on Flow Selection
[15:58:19] <nevil> .. was draft-molina psamp draft
[15:59:11] <nevil> .. is about how to select which flows you may want in a sample
[16:02:36] <nevil> Last presentation - Olav Kvittem on Flow Performance parameters (IEs)
[16:09:50] --- gerhard has left
[16:12:49] --- irino has left
[16:16:12] <nevil> End of session
[16:16:19] --- nevil has left
[16:16:21] --- andrjohn has left
[16:20:57] --- irino has joined
[16:27:22] --- irino has left
[16:35:02] --- oak has left: Disconnected
[16:58:06] --- irino has joined
[16:58:36] --- irino has left