Improving Video Delivery and Streaming Performance with...
summary
TikTok and CDN77 build an edge-first pipeline that merges CMAF chunks into unified MP4 media objects, embeds a JSON playlist, and performs dynamic URL rewriting at CDN edges to serve freshest, cacheable responses. The system generates Media First blobs on edge nodes by combining segments and prefetching, reduces origin fetches and misaligned segment requests, and achieves measurable gains including 2.5% reduction in first-frame delay and 9% reduction in stalling while improving cache hit ratios.
AI-written summary · about
chapters (9)
- 0:00TikTok live streaming challenges
- 1:30FLV and CMF for live streaming
- 3:42Performance metrics for TikTok live
- 5:10Reducing RTDs with first segments
- 6:59Segment misalignment in parallel requests
- 7:39Merging audio video into MP4
- 9:06Edge computing for media segments
- 13:45Results and improvements
- 14:51Future plans and takeaways
related talks
10mEnhancing CDN Performance and Cutting Egress Costs in Large Video Libraries
36mCHUNKY MONKEY
15mUsing The Power of Edge Compute To Enhance Streaming Delivery
20mDecentralizing Video Streaming
41m8K Codec Low-Latency Livestreaming, and AI Streaming
26mOptimizing Video Performance for Web Developers
26mBandwidth Ballet: Optimizing Video Performance for Web Maestros
16mCDN Challenges of HTTP-based Low Latency Live Streaming Delivery