jfm@cablelabs.comMinutes of the IP Flow Information eXport (IPFIX) WG
IETF 57, Vienna, Wednesday July 16, 2003
58 people in attendance
Reported by Dave Plonka & Nevil Brownlee (co-chairs) based on notes from
Greg Ruth.
The meeting agenda and slides are available here:
http://ipfix.doit.wisc.edu/IETF57/
[please see the agenda slides there for the sequence of topics:
http://i
pfix.doit.wisc.edu/IETF57/IETF57_IPFIX_agenda.ppt]
--
Juergen Quittek presented the requirements draft
"draft-ietf-ipfix-reqs-10.txt" and differences from -09 to -10.
[see slides for details:
http://ipfix.doit.wisc.edu/I
ETF57/IPFIX-reqs-IETF57.ppt]
The draft says that confidentiality SHOULD be implemented and that
anonymization MAY be. Current practice by the flow-export user
community does not depend upon these features, hence the meeting's
consensus was that the IPFIX requirements draft does not require them to be
implemented.
We discussed Allison Mankin's (Transport Area Director) suggestion that
IPFIX make confidentiality and anonymization to both be REQUIRED (MUST).
One participant noted that export restrictions (of encryption
technology) might restrict the distribution of such an
implementation.
The few issues remaining will be discussed on the mailing list to
produce the next revision in August. [If the changes continue to be minor
edits, it will not require another WG last call.]
--
Simon Leinen presented the changes to the individual draft
"draft-leinen-ipfix-eval-contrib-01.txt" and differences from -00 to -01.
[see slides for details:
http://ipfix.doit.wisc.edu/IETF57/eval.pdf]
His suggestion was that we accept this a Working Group document, go to WG
last call soon, and submit it with a request to publish it as an
Information RFC.
This proposal met with agreement amongst those in attendence.
[This would not be submitted until the requirements draft is
resubmitted to IESG]
--
Status of the four new WG drafts was reported by their respective
editors:
* Tanja Zseby presented the Applicability Statement draft.
This is now a working group draft, based on her previous
individual draft.
[see slides for details:
http://ipfix.doit.wisc.edu
/IETF57/ipfix-as-ietf57.ppt]
Reinaldo Penno presented the sub-topic of middleboxes.
[see slides for details:
http://ipfix.doit.wisc.edu
/IETF57/ipfix_middleboxes.ppt]
We discussed where this middlebox-related content should reside. It was
suggested that perhaps it should be in a seperate draft. The chairs
suggested that a "few [middlebox-related] paragraphs" should then be added to
the architecture draft and that the seperate draft be initiated as the
effort of an individual (to limit the amount of new work by the WG). A hum
showed support for this.
Juergen Quittek offered to write an individual draft.
* Ganesh Sadasivan presented the Architecture draft:
draft-ietf-ipfix-arch-01.txt
[see slides for details:
http://ipfix.doit.wisc.edu
/IETF57/IPFIX_Architecture.ppt]
Ganesh noted that the flow definition was modified to include
encapsulated IP packets. It was undecided as to whether this was an
improvement; discussion should continue in the mailing list.
It was mentioned that the Denial-of-Service section regarding
"network under attack" is unclear. It is perhaps just one case which
might invoke overload behavior. This section needs to be clarified or
reworked.
The editors asked for more review and input on this draft via the
mailing list.
* Juergen Quittek presented the Information Model draft.
[see slides for details:
http://ipfix.doit.wisc.edu
/IETF57/IPFIX-info-IETF57.ppt]
Juergen mentioned that the the definitions named with the "ipdr"
prefix should change.
XML is being used to define the information elements, which are then
parsed and rendered in an ASCII representation for the draft. The
editors have some potential improvements to this process, and the
audience didn't raise any objections to its use.
The draft proposes to provide extensibility by using separate
namespaces for sets of vendor-specific extensions. Such extensions would be
identified using IANA-defined Enterprise numbers, as was proposed in early
drafts of DIAMETER. It was noted that such an approach will require a good
way of encoding the Enterprise numbers in IPFIX templates.
Regarding the type space, the issue was raised about whether it should be
defined/duplicated within the info model draft or be defined in the Info
Model draft, with encodings specified in the Protocol draft. Nevil
believes the latter is preferable.
Discussion should continue.
* Benoit Claise presented the IPFIX protocol specification draft.
[see slides for details:
http://ipfix.doit.wisc.edu
/IETF57/IPFIX-protocol-IETF57.ppt]
It is currently just the initial version; Benoit asked that the
participants please review it and provide input [via the mailing list].
Mentioned that terminology should be consistent throughout the
drafts.
We discussed the use of "sync" packets. Jurgen suggested that sync
information should be exported in options data records. This
suggestion was well received.
--
Three short slide-shows by folks interested in IPFIX were presented.
* On behalf of Luca Deri (who could not attend), Nevil Brownlee
presenting slides on "nFlow".
[see slides for details:
http://ipfix.doit.wisc.edu
/IETF57/IETF_Vienna-nFlow.pdf]
* Chang Kim presented a slide show about "Per-packet Recond Export
Proposal (pktId)".
[see slides for details:
http://ipfix.doit.wisc.edu
/IETF57/Per-Packet_Record_E
xport_Proposal_-_ipfix.ppt]
* Maurizio Molina presented a proposal to add Flow Sampling to IPFIX.
[see slides for details:
http://ipfix.doit.wisc.edu
/IETF57/Flow_sampling_ipfix.ppt]
--
The chairs reviewed the proposed various updates to WG Milestones. No
changes were suggested so these will be submitted as listed. [see slides for
details]
--
$Id: minutes.txt,v 1.8 2003/08/11 14:41:54 dplonka Exp $
|