[15:21:18] --- LOGGING STARTED
[18:01:21] --- sleinen has joined
[18:01:34] <sleinen> Juergen Quittek is reviewing the agenda
[18:01:44] <sleinen> I-D status
[18:01:58] <sleinen> PSAM Framework: in WG last call; not enough reviews to close the last call
[18:02:35] <sleinen> Please review the documents (in WG LC) and send your comments!
[18:02:50] <sleinen> Packet selector/packet information I-D: ready for WG LC?
[18:03:04] <sleinen> other documents (3): depend on IPFIX
[18:04:02] <sleinen> "PSAMP Framework Document" - presentation by Nick Duffield
[18:04:36] <sleinen> [see slides in proceedings]
[18:05:51] <sleinen> [on encryption - mandatory to implement and optional to use]
[18:06:17] <sleinen> Andy Bierman: This is good because it allows us to leverage IPFIX without having to define a variant of it.
[18:06:37] <sleinen> Condifentiality is important for privacy reasons.
[18:07:06] <sleinen> Juergen Quittek: This issue has been settled in IPFIX in the meantime
[18:07:18] <sleinen> [on timestamp resolution]
[18:07:32] <sleinen> Nick: timestamps SHOULD have microsecond resolution
[18:07:58] <sleinen> Juergen: IPFIX definitely supports microsecond format, maybe not requirement to support microsecond precision
[18:08:14] <sleinen> PSAMP may want to be stricter and require a certain precision
[18:08:31] <sleinen> Andy Bierman: just because you have units of usec doesn't mean the clock is ticking in microseconds
[18:08:56] <sleinen> Juergen: It has been brought up in IPFIX whether NANOseconds should be supported too
[18:09:54] <sleinen> Emile ?: If we have microseconds for flows, then we might want nanoseconds for packets.
[18:10:58] <sleinen> Benoit Claise: leave liberty to the implementation about which precision has to be supported
[18:11:33] <sleinen> Randy Presuhn: as a general "future-proofing" consideration, you want to go with the finest resolution that there might be an application for.
[18:13:15] <sleinen> As long as you have a way to specify a (lower) actual precision
[18:13:30] <sleinen> Simon: The important thing is we need a way to do this
[18:13:41] <sleinen> Andy: Would a MIB object be sufficient?
[18:13:58] <sleinen> Simon: For me, yes.
[18:14:17] <sleinen> Nick: Like the export stream self-sufficient, without need to look at the MIB to interpret it.
[18:14:35] <sleinen> Juergen: This is what option records should be used for
[18:14:49] <sleinen> [Even if things like this aren't likely to change over time]
[18:15:00] <sleinen> Next topic: selecting encrypted packets
[18:15:07] <sleinen> [see slides]
[19:33:34] --- sleinen has left