Low Latency Live from a Different Vantage Point
summary
Claims a different architecture that treats transcoding and distribution as the latency root cause and replaces file I O heavy pipelines with a decentralized content fabric that does just-in-time transcoding and playlist synthesis. Uses CMAF chunked encoding delivered as HLS/DASH, yields ~3.5 second end-to-end latency, serves 200,000 segments at the top bitrate, reaches 50,000 concurrent clients, and keeps ingress/egress load modest (ingress <15–50% in tests), showing the system actually works at scale without subsegment churn.
AI-written summary · about
chapters (5)
- 0:00Start
- 0:21Background
- 2:57Idealized Chunked Encoding Transmission System
- 3:51Sources of Delays
- 5:11Side Effects
related talks
21mThe challenges of deploying Apple's Low Latency HLS In Real Life
16mCDN Challenges of HTTP-based Low Latency Live Streaming Delivery
44mLive Video in a Modern Web Environment
27mPerformance of Low-Latency DASH/CMAF and Low-Latency HLS Systems
36mCHUNKY MONKEY
41m8K Codec Low-Latency Livestreaming, and AI Streaming
10mMeasuring the live video latency at scale
35mLive Low Latency Video Streaming