Frontend Perfor
summary
Front-end work at Facebook focuses on measurable wins: end to end latency splits into 50% render, 25% network, 25% server, site speed is sampled at 1 in 10,000 page views and logged by a monitoring system, static resources are declared separately and clustered so JavaScript packages fall ~50% while delivered bytes drop, an AJAX layer called Quickling cuts client render time 40–50%, and a client page cache saves ~20% of homepage hits and speeds render another 3–4×.
AI-written summary · about
chapters (11)
- 0:00Introduction to Facebook performance
- 1:00Importance of site speed
- 2:50Unique engineering challenges
- 5:22Defining web performance
- 6:43Calie: Performance monitoring
- 9:28Analyzing performance regressions
- 11:10Technical metrics and actionable items
- 12:41Haste: Static resource management
- 20:11Quicklink: Ajaxification
- 30:17Page cache and performance results
- 37:58Conclusion and Q&A
