This week, I’m at the Internet Engineering Task Force 118 meeting in Prague, where we’re working on a new protocol: Media Over QUIC. We’re hoping this solves two converging problems when it comes to latency: Real-time collab apps want low latency, high quality, and to be able to scale a single session to a million users. The live streaming market wants to bring down latency to improve real-time interactivity. Learn more about how these two problems come together, and how we can make a difference with #MediaOverQUIC, here.
There's something else that would be super helpful that WebRTC is terrible at, namely session startup latency. Media over QUIC is suspiciously similar to Media over RTMFP, and we know how fast that started sessions back in 2010. So only 14 years later to get to the same point again?
Then be sure to put "PR-SCTP like" functionality into MediaOverQUIC (e.g., setting timed reliability parameters commensurate to the present minimum round-trip delay estimate). Definitely doable (I am PR-SCTP RFC co-author for others reading this - Cullen knows).
I love how you are still working on stuff that makes the Internet work better for everyone.
Prague, protocol, standards bodies, this all sounds vaguely familiar! 😉
Latency is not fundamentally a protocol problem; it’s an engineering problem.