Three Roads to Jerusalem
summary
Compares DASH ULL, LHLS and ALHLS and claims a single-set workflow can serve all three; prototypes a pipeline using Big Buck Bunny and ffmpeg nodejs to emit chunked CMAF segments, a manifest proxy and a simple MSE player; measures request-rate cost up to 720 requests per minute and shows h2 push cutting part transfer from ~592ms to ~252ms; documents CDN/origin gaps around byte-range push and partial-object caching; prototype works but common A/V segments delivery needs CDN and spec fixes before production.
AI-written summary · about
chapters (16)
- 0:00Intro
- 1:43Low Latency
- 4:37Religions
- 5:18Common Differences
- 9:21Chunk Transfer
- 12:18NonChunk Transfer
- 13:43Internal Segment Description
- 14:54Longer Segments
- 16:57Parts
- 19:35Request Rates
- 21:55HLS Push
- 26:35The Solution Example
- 28:55The CDN Client
- 30:20Big Buck Bunny
- 33:58Production Issues
- 35:53Conclusions
related talks
43mDASH and Media-over-QUIC Transport Face-Off: Performance Showdown at Low Latency
19mLow Latency Live from a Different Vantage Point
20mDecentralizing Video Streaming
36mCHUNKY MONKEY
24mMeet the Packets: How audio travels into your browser
58mModern & secure adaptive streaming on the Web
16mImproving Video Delivery and Streaming Performance with...
46mModern & secure adaptive streaming on the Web