Folks,
Attached are minutes for the tictoc meeting at IETF82. Thanks to Dave
Marlow for producing them. If you have any comments or corrections feel
free to send them in. Note, the deadline for changes is 5 January.
Regards,
Karen
DRAFT (23 Nov 2011)
Minutes for TICTOC meeting @ IETF-82
17 November 2011, 13:00 CST (05:00 UTC)
1. Administrivia and Agenda Bashing
The meeting started at 13:00 CST. Karen O'Donoghue chaired the meeting with
Yaakov Stein
connected via webex. Dave Marlow took minutes. Tim Plunkett was the jabber
scribe.
Karen bashed the agenda and the blue sheets were distributed. Karen provided
the status
for the Working group since the last meeting: 3 working group drafts, 6
individual
submissions, and one ad-hoc interim webex meeting.
2. ITU-T SG15/Q13 update
Stefano Ruffino provided slides for an ITU-T SG15/Q13 update which Tim Frost
presented.
SG15/Q13 had an interim meeting in September 2011. There was progress in
both the
transport of frequency in packet networks (G.826x series) and the transport of
time
in packet networks (G.827x series). Synchronous Ethernet and SDH deliver
frequency
while PTP delivers time. In answering a question from Yaakov, Tim said that
the
work in the transport of time within SG15/Q13 assumes that frequency has
stabilized,
though future work may not make this assumption. In December 2011 there will
be a
meeting of SG15 where a number of the frequency related documents and the
G.8271
time synch document are expected to be consented.
3. Precision Time Protocol Version 2 (PTPv2) Management Information Base
Tim Frost provided a PTPv2 MIB discussion based on
draft-ietf-tictoc-ptp-mib-00. This
MIB covers all of the PTPv2 devices. The MIB was in working group last call
but there
were no comments received on the list. Later there were some comments by Bert
Wijnen
and there is work going on to resolve these comments. A new draft should be
released
soon. There was a discussion as to whether some of the MIB attributes could be
writeable. Yaakov and Tim agreed that this is possible for local configuration
but performance monitoring was out of scope for this MIB. Anyone who wants
writeable attributes should send an email to the list identifying the
attributes
they want to be writeable. Tim mentioned that there is interest in a second
MIB
which would provide performance monitoring. The MIB requires a thorough "MIB
Doctor" review which may not happen until the working group completes their
review.
Karen asked for anyone who reviews this even with no comments to please post
that
they support advancing this draft to the list.
4. Transporting PTP messages (1588) over MPLS Networks
Manav Bhatia lead a discussion on Transporting PTP messages (IEEE 1588) over
MPLS
Networks, draft-ietf-tictoc-1588overmpls-02. He said that the 02 changes
revised the
draft based on the mailing list comments including fixing the Terminology
section;
removing the Entropy Label; removing 1588 over pure MPLS mode (i.e. 1588 packet
without an IP/UDP or ethernet headers); removing the part about distributing
pseudowire
labels; and a general cleanup. Distributing pseudowire Labels can be taken up
as a
different draft. Greg Mirsky had said that he sent comments to the list which
were not discussed, he was asked to resend those comments to the list which he
did. There was a discussion on MPLS Fast Reroute (FRR) in section 8 of the
draft, there is a guarantee for end-to-end protection but that the
recovery time is a "SHOULD" and not a "MUST". In discussing the next steps
Yaakov recommended reversing the proposed order of requesting outside review,
first MPLS should be satisfied and then go to the OSPF and IS-IS working
groups.
5. Time Synchronization Protocol Security Requirements
Karen briefed the newly released draft-mizrahi-tictoc-security-requirements-00.
Karen said that it will provide both a list of TICTOC and IEEE 1588 PTP security
requirements. This draft provides security threats; security requirements;
a summary of requirements; additional security implications; and issues for
further discussion. When Karen asked whether to make this a Working Group
document, Yaakov thanked the authors, indicated that it was a great start and
supported making this document a working group document. There were no
objections
to making this a working group document.
6. IPsec security for packet based synchronization
Yang Cui on behalf of the author, Yixian Xu, presented
draft-xu-tictoc-ipsec-security-for-synchronization-02. This draft has had a
large volume of discussion on list . There have been two basic questions that
have been brought up on the list: Do we need to encrypt timing packets? Do
we need to identify and decrypt timing packets right away (before decrypting
all traffic)? Yang indicated that the answer to question 1 is yes for 3GPPP
Femtocell and that the draft provided the only efficient mechanism for
carrying out a solution to the second question. It had been brought up on the
list that if timing packets were easily identified then they were more
susceptible
to attackers, for which Yang disagreed. The authors a preparing a new version
of
the draft which addresses the points discussed on the list.
7. Control Messages Protocol for Use with Network Time Protocol Version 4
Karen discussed draft-odonoghue-ntpv4-control-01. The NTP Mode 6 messages were
documented in Appendix B of RFC 1305 but this material was not incorporated into
RFC 5905 and RFC 1305 has been obsoleted. The 01 draft updated the Mode 6
messages
to reflect implementation advances since RFC 1305. This should become a Working
Group document (probably the NTP working group). A decision is needed whether
this should be informational or a standards track document, it describes
optional
features but used provides the requirements. Karen ended requesting volunteers
to
review this draft and to document NTP interleave messages and address autokey
algorithm agility.
8. Network Time Mechanisms for Improving Computer Clock Accuracy
Dave Marlow discussed draft-marlow-tictoc-computer-clock-accuracy-01. This
draft
is looking at new mechanisms beyond those identified in NTPv4 (RFC5905) that
provide increased time synchronization accuracy for operating system clocks'
time.
The ID identifies three candidate mechanisms for experimentation. In addition
to
NTP Interleaved which is in the NTP distribution but not the standard, the
draft
looks at two other mechanisms which merge the better capabilities of NTP and
IEEE 1588. The 01 version of this draft: adds security and network robustness
to performance as primary goals; NTP interleaved standards action decision
expanded beyond accuracy potential benefits; and analysis of the experiments
and those described in Dr. Mills’ new book (Network Time Synchronization—the
Network Time Protocol on Earth and in Space, Second Edition, CRC Press).
The new draft identifies a need for more data and guidance in selecting NTP
operational modes (i.e. tradeoffs between Client/Server and Interleaved
Broadcast modes, and the use of Interleaved Symmetric mode). Dave mentioned
that more than just common metrics are needed for comparing the candidate
mechanisms, a benchmarking methodology, similar to those produced by the
Benchmarking working group, appears to be the best approach. Al Morton
said that more than time synch metrics are needed, a benchmarking
methodology could identify the need to get to a stable state before
collecting data. Tim Plunkett presented his past and current experiments
on the Interleaved NTP interleaved modes as well as client/server mode. Tim
discussed his experiments and how they compared to the experiments described
in Dr. Mill's book. Dave and Tim concluded indicating that there were a number
of reasons to standardize the NTP Interleaved modes, that the other two
mechanisms should be pursued in parallel and that a time synchronization
benchmarking methodology draft should be pursued.
9. Associate PW label with PTP application
Xihua Fu discussed draft-fuxh-tictoc-associate-pw-with-ptp-00. This draft
looks at how a PTP PDU is mapped inside a PW/LSP via Ethernet encapsulation.
This draft considers the Dry Martini model in which a pseudowire is established
between two endpoints directly and no PTP LSP is needed. There was discussion
on traffic engineering which is not considered by this draft. A question that
came
up was if you have back to back pseudo wire connections without an LSP, how
would
you do traffic engineering. Someone pointed out an apparent discrepancy with
the
PDU formats (length always one byte). Yaakov questioned why label
advertisements
were being done. Manav Bhatia agreed to work with the author about these
issues.
Xihua is to send an email to the list to summarize the issues brought up on
this
draft.
10. PDV-based PTP LSP Setup, Reoptimization and Recovery
Junhui Zhang discussed draft-zhang-tictoc-pdv-lsp-00. From the ID abstract,
"This
document defines a mechanism for the setup, reoptimization and recovery of PTP
LSP
based on the PDV metrics between the 1588 Master and the 1588 Slave". The
presentation discussed the PTP communication goes through an MPLS-TE third
party network. Junhui discussed the PTP LSP setup, congestion detection, PTP
LSP reoptimization and recovery and the PTP Master recovery. There was some
concern that this is a duplicate of a draft already submitted to an MPLS
working group. Yaakov was concerned with this draft from a timing view, he
felt that the worst thing to do is to change the path for your slave, a 50%
loss
is better than changing the path where time is needed. Tim Frost said you must
know the congestion on the other path or you should not switch. Yaakov said
that
both MPLS and timing people are needed to solve this, and that he has already
sent an email to the list concerning this draft.
The meeting closed with one minute to spare.
_______________________________________________
TICTOC mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/tictoc
