Joint NTP/TICTOC Working Group Meeting
Tuesday, July 27, 2010, 1520 Ð 1700

The jabber log can be found at: 
http://jabber.ietf.org/logs/tictoc/2010-07-27.txt 

The meeting was called to order by co-chairs Yaakov Stein and 
Karen O'Donoghue. Sterling Knickerbocker took the minutes, and 
Brian Haberman acted as jabber scribe. Blue sheets were 
distributed, and the note well warning presented. 

The proposed agenda was:
1. NTP issues
   - Karen
2. 1588 over MPLS 
   - Yaakov  
   - Shahram   
   - Ron Cohen  
   - Lizhong Jin (P2MP LSP with co-routed reverse path)
3. Timing Security
   - Stefano
   - Rock   
4. Timing Management
   - Tim Frost
5. Other
   - Stefano (ITU SG15 Q13 status update) 

NTP issues
==========
Slides:
www.ietf.org/proceedings/78/slides/tictoc-10/tictoc-10.htm

Karen O'Donoghue reviewed recent accomplishments in the NTP WG. 
The four primary documents have been published as RFCs (RFC5905, 
RFC5906, RFC5907, RFC5908). There is a new draft under 
discussion on list (draft-chen-ntps-ra-opt-00.txt). There is 
some concern about whether this option should be supported at 
all. Further discussion is requested on the mailing list, and 
guidance has been solicited from the 6man working group chairs. 

There has been some discussion regarding moving the core NTP RFC 
further along the IETF standards track because of its level of 
maturity. Further discussion will be held off until after the 
Administrative Plenary due to the possible changes to that 
process. Yaakov commented that discussions have taken place 
about making exceptions for standards track documents and the 
multiple implementation requirement. 

ITU SG15 Q13 status update 
==========================
Slides: 
www.ietf.org/proceedings/78/slides/tictoc-4/tictoc-4.htm

Stefano Ruffini presented an update on the activities of ITU-T 
SG15 Q13 which has been in operation since 2004. Several 
recommendations have been completed including G.8261, G.8262, 
and G.8264. An IEEE 1588 Frequency profile is under development 
and will be followed by a Time of Day profile. Additional 
details are available in the slides. 

Danny Mayer asked why there were no IPv6 based profiles under 
development in the ITU-T. Stefano indicated that it could be 
could be a future work item. Peter Lothberg asked which version 
of UTC is being specified. Stefano responded that it is the most 
local UTC, and Peter commented that several companies may have 
several UTC sources that may not match within the desired 
performance window. 

1588 over MPLS 
==============

There were four presentations related to IEEE 1588 over MPLS. 

1588 MPLS encapsulation
-----------------------
Slides: 
www.ietf.org/proceedings/78/slides/tictoc-0/tictoc-0.htm

Yaakov Stein presented a summary of the 1588 MPLS encapsulations 
options discussed at IETF77 and discussed what would need to be 
done next. There was a fair amount of discussion related to the 
basic requirements associated with this effort. One question 
asked was why do you need special treatment for timing packets. 
We need to more clearly articulate the requirements and see if 
there are already tools in MPLS to address them. Putting 
timestamps as close to the hardware as possible has already been 
solved. If a Boundary Clock (BC) or Transparent Clock (TC) is 
used, work will be required to handle the signaling. How is 
packet routing/rerouting handled? Luca Martini commented that we 
need to define the service of the network.  A transparent clock 
will only work on a known network.  Define an MPLS label so the 
network knows it is a timing packet and what to do with it. The 
requirement for pursuing a special encapsulation is not clear, 
and the specifications required for the encapsulation are not 
developed. 

1588 over MPLS
--------------
Slides: 
www.ietf.org/proceedings/78/slides/tictoc-5/tictoc-5.htm

Yaakov Stein presented a set of slides for Shahram Davari who 
was unable to attend. These slides propose an approach to IEEE 
1588 encapsulation. Luca Martini commented that this is fine 
with respect to encapsulation but still needs some signaling 
work. Something needs to be updated so the router can recognize 
this as a timing packet. Another question was does the router 
need to recognize it as a timing packet? Does 1588 support P2MP 
LSPs? It appears the answer is yes for both Ethernet and IP. 
Peter commented that you need to make sure you do not hard code 
the solution. There was a comment that TCs are not needed; 
however, Steffano commented that TCs are needed for very 
accurate time sync. Mark Glasser commented that we should 
reorder the options because some are harder/more work. This 
appears to be an easier solution.

(Direct) PTP over MPLS
----------------------
Slides: 
www.ietf.org/proceedings/78/slides/tictoc-8/tictoc-8.htm
Draft: 
draft-ronc-ptp-mps-00.txt (expired)

Ron Cohen provided an overview of his expired draft on PTP over 
MPLS providing a justification for PTP directly over MPLS. Peter 
Lothberg asked how big is the MPLS cloud in km or miles?  If you 
make it large, say 500km of fiber, the delay variation will be 
on the order of 200ns. Craig commented that it is assumed that 
you can synchronize the boundaries of the MPLS cloud, but that 
is the basis of the problem Ð how do we sync the clocks on the 
edge of the cloud? Craig asked if we have looked at the effects 
on time sync if the service being provided may not have the same 
time reference in the LSR when supporting multiple clocks. 
Further discussion was moved to the mailing list. 

P2MP LSP with co-routed reverse path
------------------------------------
Slides: 
www.ietf.org/proceedings/78/slides/tictoc-1/tictoc-1.htm

Lizhong Jin presented a set of slides on P2MP LSP with co-routed 
reverse path. Yaakov commented that trying to force two slaves 
to send their delay_req at different times to avoid congestion 
at the Grandmaster sounds complicated. Further discussion was 
moved to the mailing list. 

Timing Security
===============

There were two presentations on timing security. 

Stefano Ruffini: Packet Timing Security Aspects.
------------------------------------------------
Slides: 
www.ietf.org/proceedings/78/slides/tictoc-2/tictoc-2.htm

Stefano provided a discussion on some thoughts on basic 
architecture and requirements with respect to timing and 
security and possible ITU-T efforts. This is one area where the 
ITU-T would like to be able to leverage work done in the IETF.  

Security Requirement for 1588v2 over IPSec
------------------------------------------
Slides: 
www.ietf.org/proceedings/78/slides/tictoc-3/tictoc-3.htm

Rock presented two proposed approaches in 3GPP and ITU for IEEE 
1588v2 protection using IPsec, entire protection and partial 
protection. In the entire protection approach, all of the 1588v2 
packets are protected with IPsec. For partial protections, the 
general messages are protected while the event messages are not. 
There was concern expressed that encryption of timing packets 
would impact the synchronization accuracy. Further discussion 
has highlighted the difference between encrypting timing packets 
to secure them versus having to compensate for transmission of 
timing packets over a link that is encrypted. Further discussion 
on the topic is required. It was also unclear whether this work 
should be pursued in the TICTOC or IPSecME working groups. 

Timing Management
=================
Drafts:
draft-frost-tictoc-management-00.txt
draft-frost-tictoc-ptp-slave-mib-00.txt

Slides: 
www.ietf.org/proceedings/78/slides/tictoc-6/tictoc-6.htm
www.ietf.org/proceedings/78/slides/tictoc-7/tictoc-7.htm

Tim Frost presented two drafts on network management. Due to 
time constraints, discussion was moved to the mailing list. The 
chairs thanked Tim for the submission of drafts to initiate the 
conversation. 

Karen wrapped up meeting with a reminder to attendees to ask 
your questions on the mailing list as some may have been missed 
during the meeting. The chairs also indicated that they plan to 
hold a series of conference calls to progress to work before the 
next meeting. The chairs thanked everyone for the contributions 
and discussion. 

The meeting was adjourned at 17:05 pm.