Hi Jinghao Yuan. 1. What would happen with CUBIC and/or old QUIC flows, if new QUIC flow would get an advantage? Will they get the much less performance? CC fairness was discussed many times. 2. I am not concerned about privacy that you have discussed in the security section. I am concerned that the client may abuse this mechanism. For example, start always 25% faster than recommended from the server to gain advantage even against new QUIC flows. It has been proven many times (for example, HPCC) that telemetry gives the big advantage for CC. But it was for “closed domain”. The Internet is not a trusted environment. Do you restrict the scope for “closed domain”? Eduard ________________________________ From: 袁靖昊 <[email protected]<mailto:[email protected]>> Sent: Sunday, October 19, 2025 10:47 AM To: [email protected]<mailto:[email protected]> <[email protected]<mailto:[email protected]>> Subject: New ID--Exchanging Congestion Control Data in QUIC
Hello colleagues at QUIC WG. I'm Yuan Jinghao from ByteDance, and I'd like to introduce to you a draft we recently submitted. We have noticed that in various scenarios, there is a demand to represent and consume congestion control status. Based on this, we have constructed a custom QUIC frame, which is composed of multiple Network Statistics. Every "Network Statistics" records network signals such as bandwidth, packet loss, and RTT, etc. This extension allows peers to exchange their cc status and enables the original data to be echoed back to the source in a secure form. We have applied the prototype of this draft in live streaming for several years and achieved actual QoE benefits. We hope that this extension can not only be applied in TikTok live but also receive professional evaluations and guidance from the industry. Thanks, Jinghao Yuan draft-00: https://datatracker.ietf.org/doc/draft-yuan-quic-congestion-data/
