[P4-Apps] WG meeting notes, 12/19/2019

[Jeongkeun Lee]

Attendees: 

Ramesh S. (Cisco), Mukesh H. (VMware), Micheal O. (Intel), Mickey S. (Barefoot/Intel), JK L. (Barefoot/Intel).

We mostly talked about the switch behavior upon MTU exceed.

Currently the INT spec allows a transit switch that hits MTU limit to generate a report for already-collected upstream data, and start a new INT stack to gather metadata from itself and downstream switches. W/ this, a remote collector may receive report messages from two devices (the transit device and the final sink) or more devices and should collate them to construct a complete e2e data. 

Ramesh proposed to add a bit in the telemetry report to indicate collectors that this report is partial and needs to be collated.

AI: Ramesh will create a PR against the report format.

Second topic was about v2.0 timeline.

We freeze the features and changes towards v2.0, and decided to finalize the PRs by Jan 9th. We may take editorial changes for a week and cut v2.0 specs (both INT and report format) in the next meeting, Jan 16th.

AI: everyone to review the open PRs, address comments on open PRs by 1/9/2020.

Towards the end of the meeting, Michael asked a question about different but similar specs -- P4.org INT, IETF IOAM, etc. We will discuss more about this topic and the direction of p4 INT, early next year.

Thanks,

JK Lee

Mukesh Hira

[Jeongkeun Lee]

Dear members,

As indicated in the last meeting notes (below), we will have the next WG meeting on 1/16/2020 this Thursday.

Agenda items:

- close PRs and cut v2.0 specs

- overall next steps of the Apps WG

  Telemetry: relation with IETF, NIC-driven use cases and telemetry for closed-loop control

  other apps than telemetry: AI acceleration, congestion control..

Location: Barefoot Networks

Meeting time: 10 am to 11:30 am PT Remote attendees may join at: https://meet.google.com/rwy-typy-yxu

Google calendar link is here.

Thank you.

JK Lee

Mukesh Hira