On Jul 8, 2014, at 11:50 AM, Tatsuhiro Tsujikawa <tatsuhiro.t@gmail.com> wrote: > ... > Why do we have to use same frame size for HTTP/2 with video's? I'm not sure I understand the question. > We can deliver multiple DATA frames for 1 video frame. Also big frame really hurts multiplexing many people stated earlier. The question is delivering those multiple frames to ensure smooth playback of the video. In this case, using a slightly larger frame will ensure that each frame contains a full video frame (or the rest that started in the previous DATA frame). This is a balancing act with priority, which can also hurt multiplexing. At least with (slightly) larger frame sizes you can "atomically" transfer a full video frame and its associated audio, making it unnecessary to crank up the priority just to ensure you get the data when you need it - smooth and steady is what you want for video, not high volume bursts. Let's also not forget that the proposal on the table is NOT to allow an endpoint to unilaterally decide to send a 1GB DATA frame, to the detriment of all. Instead, it allows the receiver to tell the sender what its limit is (based on resources, available bandwidth, or any other criteria), which then allows the sender to adapt accordingly. _________________________________________________________ Michael Sweet, Senior Printing System Engineer, PWG Chair
Received on Tuesday, 8 July 2014 16:25:16 UTC