Distributed Request Tracing for the Streaming Video Ecosystem
summary
Project builds a working POC that extracts trace context from players and CDN logs, stitches them into request-scoped traces and feeds Zipkin for root-cause work. It instruments Dash.js and THEOplayer, uses query-string CMCD with CMCD session ID (SID) and custom org.svalabs.rid, pulls Fastly/AWS CloudFront/S3 logs into Datazoom and Splunk, and demonstrates visible alignment of player events, CDN log lines and stalls under slow 3G. Next steps are scale tests, simulated failures, more CDNs and Zipkin visualisations.
AI-written summary · about
chapters (12)
- 0:00Start
- 0:14The Four Signals of the Streaming Apocalypse
- 1:00Where in the workflow is the root cause
- 3:21Josh Evans
- 4:27Phase 1 (VOD) - Goals and Objectives
- 5:26Phase 1 Configuration - VOD
- 5:53Services & Technologies
- 6:29Unlocking Observability
- 7:54Common Media Client Data (CMCD)
- 11:06Working List: CDN Trace Context
- 16:28Phase 1 (VOD) - Next Steps
- 17:47Common Media Server Data
related talks
29mDistributed Request Tracing
50mDistributed Tracing in the Wild
24mDistributed Tracing : New DevOps Foundation.
36mDistributed Tracing: From Theory to Practice
30mDistributed Tracing: From Theory to Practice
42mAn introduction to distributed tracing and Zipkin
57mDistributed Tracing: How the Pros Debug Concurrent and Distributed Systems
15mTRACING AKKA STREAMS