Real User Performance Monitoring at Netflix scale
summary
Describes Netflix’s in-house Real User Monitoring (RUM) system that measures app performance across thousands of device types and global network conditions. Covers challenges of building a consistent cross-device RUM pipeline and how Netflix collects and processes the large-scale telemetry using open-source tools like Druid and Apache Spark, plus internal tooling to help engineers analyze user-perceived performance and guide optimizations.
AI-written summary · about
chapters (12)
- 0:00RUM at Netflix
- 4:27Netflix scale and challenges
- 8:02Performance visibility on devices
- 13:57Defining response time metrics
- 19:14Data collection implementation
- 23:13Large-scale data processing
- 28:09Efficient aggregation with sketches
- 32:06Metric visualization and analysis
- 35:57Automatic anomaly detection
- 38:45Frequent metrics and histograms
- 43:42Network models and user experience
- 47:56Summary and conclusion
related talks
16mHome Brewing R.U.M - Analyzing application performance with real user monitoring
02mMobile Real User Monitoring
05mHow real is real-user measurement?
46mNoise Canceling RUM
07mEnd User Experience Monitoring
11mAre They Really Using It? Monitoring Digital Experience To Determine Feature Effect
31mMastering Near-Real-Time Telemetry and Big Data: Invaluable Superpowers for Ordinary SREs
34mWhy Google’s CrUX Results Are Not Reproducible With Your Real-User Monitoring