De : Kuhn Nicolas <[email protected]> Envoyé : mardi 8 juin 2021 09:07 À : IETF QUIC WG <[email protected]>; iccrg IRTF list <[email protected]> Cc : Christian Huitema <[email protected]>; 'Gorry (erg)' <[email protected]>; Tom Jones <[email protected]>; STEPHAN Emile INNOV/NET <[email protected]>; Kuhn Nicolas <[email protected]> Objet : [QUIC 0RTT BDP] Update of draft-kuhn-quic-0rtt-bdp
Hi, The 0RTT-BDP draft proposes methods that improve the performance of ingress traffic for resuming sessions. In short, when resuming a session, a server might exploit the CWND of a previous connection to adapt the new session transmission rate. Following Gorry's presentation of this draft at ICCRG's IETF110 meeting, we have received many feedbacks that this updated version (we hope!) integrates. See updated version here : https://datatracker.ietf.org/doc/draft-kuhn-quic-0rtt-bdp/ In particular, this update : - Includes a new author - Clarifies the rationale behind safety guidelines when "getting-up-to speed" - Assesses a trade-off between three distinct implementation methods in QUIC - Describes three implementations methods, two of them being implemented in picoquic : o Local storage of previous BDP parameters : https://github.com/private-octopus/picoquic/pull/1204 o Using the BDP Frame extension : https://github.com/private-octopus/picoquic/pull/1209 Evaluations on the different methods are on-going and we would be pleased to present the results at IETF111. In the meantime, please let us know if you have any view on this version of the draft through email, issues (https://github.com/NicoKos/QUIC_HIGH_BDP/issues), letters, RFC2945, etc. Kind regards, Nico - on the behalf of the author team
