Session Date/Time: 17 Jun 2025 13:00

NTP

Summary

The NTP Working Group met to review recent progress, including the publication of several RFCs and the conclusion of the NTP-Tictoc WG. Key discussions centered on advancing the Rough Time draft to the IESG, progress and unresolved issues for NTPv5, and an update on 1588 (PTP) security and protocol evolution. A significant portion of the NTPv5 discussion focused on the decision to separate the time synchronization algorithms from the core protocol specification, leading to a planned consensus call on the mailing list. Plans for a hackathon were also discussed, with the next target being Montreal in November.

Key Discussion Points

Administrative & Notewell: The IETF Notewell applies. Attendees were reminded to use the Meetecho Q tool and contribute to online notes.
Working Group Status Update:
- RFC 9748 (NTP Registries) published in February. Designated Experts (Marislav, David, Kristoff) named. Registry is now operational.
- RFC 9769 (Interleaved Modes) published in May 2025.
- RFC 9760 (Enterprise Profile) published in May 2025.
- The NTP-Tictoc Working Group has concluded all its work items and has been officially concluded. Its mailing list is archived.
- The NTP Working Group charter was expanded to include 1588-related work within the IETF.
- NTP over PTP is progressing to the IESG.
Rough Time (draft-ietf-ntp-rough-time):
- The Working Group Last Call feedback has been addressed by the authors, including important security issues.
- The draft is considered ready to progress to the IESG.
NTPv5 (draft-ietf-ntp-ntpv5):
- Tal improved the abstract, introduction, and basic concepts sections.
- A technical change updated the poll value to zero to minimize request information.
- References to the NTPv4 interleave modes (now RFC 9769) have been updated.
- Tal noted that most requirements from the NTPv5 requirements document are aligned, with two remaining open issues discussed on the mailing list.
- Discussion on Algorithms:
  - The chair noted a need for a formal consensus call regarding the specification of time synchronization algorithms, particularly concerning Harland's discussion on the mailing list.
  - Marcus commented that while specifying algorithms has merit, real-world implementations will vary, and it might be better to allow implementations to declare the algorithm used.
  - Ira suggested separating the base specification from algorithm documents for agility, referencing SUIT and post-quantum considerations. This would allow algorithms to be updated independently.
  - Karen recalled the original intent for NTPv4 (RFC 15905) was to separate protocol from algorithms, but Dave Mills merged them. PTP's flexibility benefits from separation.
  - Mirislav expressed concern that specifying algorithms could slow down progress and suggested implementations could initially use NTPv4 algorithms as a default.
  - Kristoff agreed that the on-wire protocol is valuable independently and supported referencing NTPv4 algorithms as a default if needed.
  - Tal emphasized separating algorithms from the protocol for faster progress and the likelihood of multiple algorithms in practice.
  - Ira clarified that the discussion was about time synchronization algorithms, not security algorithms, though the community might be more accustomed to multiple security algorithm specifications.
  - Kristoff reiterated strong support for separating the on-wire protocol from clock-setting algorithms, to avoid publication delays.
  - Mirislav noted PTP separates core protocol from synchronization algorithms, which are often in profiles. He argued that client-side algorithm implementation issues primarily affect only that client.
  - Tal noted that PTP's default profile does not typically include algorithmic aspects, which are often left to specific implementations.
  - The chair stressed the need to formalize the decision on algorithm separation given the history of recurring issues.
1588 Update (PTP):
- The 1588 security working group has been working to resolve vulnerabilities identified by David Venhook and is moving towards consensus on improvements to the base specification.
- Martin updated on ongoing discussions for NTS for PTP, focusing on replay protection and comparing different solutions.
- The FAR (Formal Action Request) for the 1588 security work is before NESCOM this week for approval.
- CSPTP (Client-Server PTP) work is ongoing, with an initial proposal under development following a decision on the approach.
- A "roll-up" of all existing 1588 amendments (since the 2019 publication) is being undertaken before further amendments can be made.
Hackathon: A Way Forward:
- The planned hackathon in Madrid might not proceed due to lack of NTS/NTPv5 implementation readiness or in-person attendance.
- Remote participation is possible and has been utilized previously (e.g., Mirislav).
- The group aims to target a hackathon in Montreal in November.
- A call was made for anyone implementing NTS for NTP or NTPv5 to get in touch with the chairs. Martin indicated his NTS for PTP implementation is not yet ready for testing.

Decisions and Action Items

Decision: The draft-ietf-ntp-rough-time document is ready to be progressed to the IESG.
Decision: The NTP-Tictoc Working Group has concluded its work and been closed.
Action Item (Chairs): The chairs will initiate a consensus call on the mailing list regarding the separation of NTPv5 protocol specification from time synchronization algorithms.
Action Item (Chairs): Check with David on his potential availability for a hackathon.
Action Item (All): If aware of anyone implementing NTS for NTP or NTPv5, please put them in touch with the chairs.

Next Steps

NTPv5: Engage in the upcoming mailing list consensus call regarding algorithm specification and separation.
1588 (PTP): Continue discussions on replay protection and NTS for PTP. Monitor the NESCOM approval of the 1588 security FAR.
Hackathon: Plan for a hackathon activity targeting the Montreal IETF meeting in November, focusing on NTPv5 and NTS if sufficient interest and implementation progress emerge.
Next Meeting: The next NTP WG meeting will be held in the context of the Madrid IETF meeting. The schedule will be communicated once finalized.