Transport Protocol Performance over Deployed Satellite Systems

WEDNESDAY 26 July 9:30-10:50 Golden Gate 4

This meeting is for participants to discuss transport protocol performance over deployed satellite systems. The meeting is related to activities discussed on the ETOSAT mailing list, associated with PANRG.

Organiser Gorry Fairhurst - please contact via email to add agenda items

Remote participants please use: https://meet.jit.si/IETF117-satellite-perf

Agenda (Timing to be confirmed)


IETF side meetings are open for any registered IETF participant to attend and are subject to the IETF Note Well.

Brief Summary

A group of people met in SF as a side meeting. There were approx 40 people, some on-line. The IETF notewell was used. We discussed satellite systems, and also saw data from Joerg on Hystart/Hystart++ pathologies over LEO and GEO satellite. Picoquic, quiche, and Linux TCP were mentioned, and the group agreed to catalogue the requirements for satellite systems with respect to exit of slow start to avoid overshoot, but also to avoid overly conservative early exit. There were suggestions that the group might be able to propose a solution - preferably not specific to satellite systems (so might also help for other paths with high delay or high delay variation), but which would benefit a broad range of satellite systems.

At the meeting it was decided to continue discussion on the ETOSAT mailing list and also to open a github repository to cpature issues, proposed text etc. the plan at this moment is to have one (possibly two) drafts for the next IETF - likely targeting TCPM as a proposal for Hystart++, or for CC WG if that was thought appropriate. It might also be possible to offer more data in ICCRG or another side meeting.