[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

draft minutes of Seoul session

Dear all,

Below find the draft minutes of our session in Seoul.

Please read and comment.


Juergen Quittek        quittek@ccrle.nec.de        Tel: +49 6221 90511-15
NEC Europe Ltd.,       Network Laboratories        Fax: +49 6221 90511-55
Kurfuersten-Anlage 36, 69115 Heidelberg, Germany   http://www.ccrle.nec.de

Minutes of the PSAMP session at IETF 59
Tuesday March 2, 09:30 h - 11:30 h

Session summary

The PSAMP WG discussed the first two documents that are close
to completion (expected in April).  The framework document is
in last call and just needs some serious detailed review.
The packet selection document is close to be ready for last call.
The only major open issue is the selection of a mandatory hash
function.  The existing suggestions were technically evaluated,
but IPR issues are still unclear.

The remaining documents (protocol, information model, and MIB
module) are progressing slowly.  All of them depend on documents
of the IPFIX WG.  The editors were advised to use the time until
the IPFIX documents are stable for solving all independent

---------- 1. WG Status (Juergen Quittek)

Juergen presented the list of WG documents and summarized their
respective states.  The PSAMP Framework document is in WG last call
and still waiting for further review.  The document on Packet
Selector and Packet Information is almost ready for WG last call.
The remaining three documents (Report Format and Report Stream
Format, Export Protocol and Requirements for Collectors, PSAMP MIB)
all depend on the IPFIX protocol and information model, which are
still under development.  Therefore progress there is slow, just
IPFIX-independent issues are progressing.

---------- 2. PSAMP Framework, draft-ietf-psamp-framework-05.txt (Nick Duffield)

Nick first reported on changes in the framework since the last session.
Confidentiality protection available for exported packet records was
required for IPFIX by the IESG.  Since PSAMP uses the IPFIX protocol,
this also applies to PSAMP.  In PSAMP confidentiality is even more
important because laso payload of packets may be exported.
Timestamps SHOULD have microseconds resolution. Accuracy of timestamps
MUST be reported.
[Benoit Claise] IPFIX already requests microseconds precision
[Randy Presuhn] Use the finest precision considered in order to be
[Andy Bierman] Is applied resolution fine to be reported in the MIB.
[Nick] It would be desirable to have the PSMAP data stream self-contained.
[Juergen] to be done by option record.
A discussion of packet selection of encrypted packet was added to the
[Nick] MUST be configurable to ignore encrypted packets.
[Andy] This is an issue for the MIB module.
The section discussing sequence numbers was extended as well as the
section on expoert timelyness.
[Nick] include sequence number in every report.
[Juergen] Could this be a matter of configuration.
[Andy] This puts a lot of additional effort.  Let's investigate it first.

Nick asked about recent developments in IPFIX. For IPFIX, SCTP is a MUST.
Does this mean SCTP-PR is required?  Is it sufficient to just support
SCTP-PR?  These questions were forwarded to the IPFIX WG session.

[Nick] Is it OK with IPFIX if a PSAMP probe just reports packet slices?
[Juergen] Yes, IPFIX supports that.  This should be explicitly mentioned
by the protocol document.

[Nick] PSAMP proble SHOULD be able to also filter non-IP packets.
[Juergen] IPFIX does not cover non-IP packets.
[Andy] we should use MAY instead of SHOULD.
[Nick] MAY is probably OK.
[Emile] It is not clear how the configuration for sampling non-IP
packets is done and how to do hash-based trajectory sampling with non-IP
[Andy] Better avoid it because of its complexity that was discovered by
the RMON WG.

There is no obvious need to delay this document until work in the IPFIX
WG has progressed further.  Therefore, this doczument can be passed to
the AD as soon as WG last call is closed.  Yet, to be discussed is the
targeted RFC status. SHOULD it be informational or standards track?
[Juergen] It's OK if requirements are implemented by standard docs
[Randy] For a framework document the statement of compliance becomes
[David Kessens] Let's discuss it later and come back to the issue on
the mailing list.

---------- 3. Packet Selection, draft-ietf-psamp-sample-tech-04.txt (Tanja Zseby)

Tanja presented changes to the Packet Selection I-D since the last
meeting.  Two mandatory hash functions were included, one for random
sampling and one for packet identifier genration.

An evalaution performed by Maurizio Molina and Nick led to the selection
of the IP Shift-XOR algorithm for random sampling and the CRC32 for
packet ID generation.

The I-D is almost ready for WG last call, it just needs one more revision.

---------- 4. Hash Function Comparison (Nick)

Nick presented slides on the comparison of hash functions prepared
by Maurizio.  They comparison covered the technical differences.
[Andy] IPR need to be included in the evaluation.
[Nick] IPSX may be protected by IPR.
[Juergen] We need to evaluate the IPR issue for all four candidates.
[Stewart Bryant] Hash function selection should also cover the IP
version IPSX does not cover IPv4.

---------- 5. PSAMP Protocol Specification, draft-ietf-psamp-protocol-01.txt (Benoit)

Benoit reported on recent changes of the protocol specification.
Most important is the fact that now the decision about SCTP as mandatory
transport protocol was made.

There were two major open issues.  First, do we want to distinguish
export with one packet from a PSAMP export?
[Juergen] We must be able to distinguish IPFIX flow records containing
a single packet from a PSAMP report on a sampled packet
[Benoit] the content is not different
[Andy] There is a difference between reporting flows and reporting packets

The seond open issue is concerned with reporting the sequential order of
sampling and filtering.
[Tanja] The packet selection draft gives directions.
[Andy] We need to make sure that option records and MIB content are

---------- 6. PSAMP Inforamtion Model, draft-ietf-psamp-info-01.txt (Juergen)

Juergen presented slides prepared by Thomas Dietz.  The major change
was the alignent with the new structure of the IPFIX information model:
The XML prepresentation of the information model was changed.  The
information model is now better aligned with the packet selection I-D.
Still open is the info model for packet filtering, for the sequence
of sampling and filtering and for the observation point.  Also, the
numbering scheme of the PSAMP fields to be registered by IANA is not

---------- 7. PSAMP MIB, draft-ietf-psamp-mib-02.txt (Juergen)

Juergen presented slides prepared by Thomas Dietz.  Thomas updated
the sampling methods according to the new version of the packet
selection I-D.  He added a simple structure for representing templates.
This needs to be further elaborated.  Most still open issues depend on
progress in the information model.

---------- 8. WG Schedule

The WG is far behind its planned schedule.  One reason for this is the
dependency on progress in the IPFIX working group.  The framework and
packet selection documents are expected to be completed soon, the other
three documents are expected to be completed one IETF meeting after the
IPFIX protocol and information model documents are complete.

-- to unsubscribe send a message to psamp-request@ops.ietf.org with the word 'unsubscribe' in a single line as the message text body. archive: <http://ops.ietf.org/lists/psamp/>