**Session Date/Time:** 07 Mar 2023 13:00

# [DETNET](../wg/detnet.html)

## Summary

This interim meeting focused on establishing the process for enhancing the DETNET data plane, with an initial emphasis on queuing mechanisms. Key updates were provided on the `draft-ietf-detnet-scaling-requirements`, and a tentative approach for evaluating solutions against these requirements was presented. Logistics for a series of open working meetings were discussed and decided upon, alongside a proposal for an informal side meeting at IETF 116 in Yokohama.

## Key Discussion Points

*   **Meeting Introduction and Announcements**: Lou Berger (Co-Chair) opened the meeting, welcoming participants and reminding them of IETF policies (Note Well, virtual blue sheet). The primary purpose of this interim was to discuss the process for data plane enhancements, not solutions, and to gather input on the plan for upcoming open working meetings. David Black (Technical Advisor) was introduced as the chair for these open meetings.
*   **Requirements Draft Update (draft-ietf-detnet-scaling-requirements-01)**:
    *   Pang Liu presented updates to the draft, reflecting comments from the adoption call.
    *   The name was changed from "large scale" to "scaling" DETNET, and corresponding text was adjusted throughout the document.
    *   The abstract's terminology was refined (e.g., "long hops per hop time variation" to "high variation in latency and hop counts").
    *   Section 3 was restructured, adding explicit requirements for flow fluctuation and clarifying relations to key attributes.
    *   Discussion points included the potential need to define "scaling" and the mapping of data plane enhancement requirements (Section 4) to technical requirements (Section 3). Specifically, it was noted that requirements 4.3 and 4.4, concerning IPv6 and MPLS data plane solutions, lacked clear one-to-one alignment and might need further refinement or a new requirement in Section 3.
    *   Authors welcome further comments on the mailing list to refine the document.
*   **Proposed Process for Data Plane Enhancements**:
    *   David Black outlined the tentative approach for informal open working meetings, with an initial focus on **queuing**.
    *   **Goals**: Establish common information and definitions, and assess criteria for different mechanisms.
    *   **Tentative Approach**:
        1.  Refine the `scaling-requirements` draft, potentially removing requirements not directly related to scaling (e.g., 4.3, 4.4).
        2.  Map proposed solutions to requirements using a 2D matrix.
        3.  Prioritize requirements into "Must," "Should," and "May" categories based on their importance for a scalable DETNET.
        4.  Converge on a set of solutions that satisfy the most important requirements and seek working group rough consensus.
    *   It was noted that these steps would occur concurrently, with feedback loops.
*   **Logistics for Open Working Meetings**:
    *   Recognizing the challenge of finding a convenient time across all major time zones, a proposal was made to **alternate meetings every two weeks** between an evening ET slot (8-10 PM ET) and a morning ET slot (8-10 AM ET). This aims to distribute inconvenience.
    *   The first two proposed meetings were April 11th (evening ET) and April 26th (morning ET).
    *   Meetings will be recorded.
    *   A participant suggested re-evaluating the alternating schedule after a few months.
    *   A participant proposed an **informal, self-organizing side meeting at IETF 116 in Yokohama** to leverage face-to-face discussions, with a focus on bringing outcomes back to the mailing list for broader input.
*   **Scope and Criteria for Queuing Solutions**:
    *   A participant (Torless) emphasized the need to define the type of goals queuing can address (e.g., latency, jitter) and consider hardware feasibility aspects for comparison.
    *   Another participant (Sheena Song) highlighted the importance of **encapsulation** as a key consideration alongside queuing mechanisms, suggesting that evaluation should include whether a mechanism requires new encapsulation. The chairs agreed that mechanisms' additional data requirements are a valid evaluation criterion.
    *   A participant (Yakov) raised **scalability concerning network load and capacity utilization**, noting that different queuing mechanisms have varying impacts on usable capacity (e.g., strict time-slotting vs. more flexible approaches). This was suggested as a crucial point for comparison and potential refinement in the requirements draft.
    *   Discussion arose about the **definition of "deterministic"** and whether DETNET aims for absolute latency bounds or allows for statistical guarantees (e.g., 99% of packets meet a bound). While acknowledging some applications require hard bounds, there was an expressed preference to explore a broader range of queuing mechanisms, including those that might offer statistical bounds, rather than imposing hard real-time requirements upfront. This is a critical point for the `scaling-requirements` draft.

## Decisions and Action Items

*   **Decision**: Informal open working meetings for DETNET data plane enhancements will proceed, with an initial focus on queuing.
*   **Decision**: The proposed alternating schedule for open working meetings (8-10 PM ET and 8-10 AM ET, every two weeks, starting April 11th/12th) was adopted. This schedule will be reviewed after a few months.
*   **Action Item**: Pang Liu and co-authors to address remaining comments on `draft-ietf-detnet-scaling-requirements-01` on the mailing list and further refine the mapping between Section 4 and Section 3, particularly concerning IPv6 and MPLS data plane solutions.
*   **Action Item**: Sheena Song will coordinate an informal, self-organizing DETNET side meeting in Yokohama, encouraging discussion on remote participation options on the mailing list for those not attending in person.
*   **Action Item**: Participants are encouraged to bring suggestions to the mailing list for refining the `scaling-requirements` draft, particularly regarding load/capacity considerations and the precise definition of "deterministic" and latency bounds (hard vs. statistical).
*   **Action Item**: The open working meetings will evaluate proposed queuing mechanisms, specifically considering their requirements for additional data or changes to encapsulation.

## Next Steps

*   The first open working meeting is scheduled for April 11th (evening ET) / April 12th (morning ET).
*   The second open working meeting is scheduled for April 26th (morning ET).
*   Further discussions on the `scaling-requirements` draft and the overall process for data plane enhancements will continue on the DETNET mailing list.
*   Organization of the informal Yokohama side meeting will commence.