Fawkes: Faster Mobile Page Loads via App Inspired Static Templating
summary
Fawkes extracts static templates from multiple HTML versions offline, serves those templates immediately from cache, and streams small dynamic patches that transform the template into the current page; a client-side JS shim enforces view invariants so scripts see the same DOM state as a normal load. Implementation uses tree-matching plus a breadth-first heuristic to cut patch generation time by orders of magnitude (≈1% more operations), and on 500 Alexa pages shows median Speed Index and TTFP improvements of ~46% and ~64% in warm caches (24% and 62% cold).
AI-written summary · about
chapters (16)
- 0:00Start
- 0:47Web Optimizations
- 1:08Fetching the Top-level HTML
- 2:15Mobile Apps vs. Mobile Web
- 3:44Mobile App Startup
- 4:39Web Page Startup
- 5:59App-like Templating for the Web?
- 6:44Fawkes
- 8:54Research Questions
- 9:31Static Template Generator
- 13:29Dynamic Patch Generation
- 15:53Challenge: Correctness
- 17:19Evaluation Results
- 17:37Warm Cache Improvements
- 18:26Cold Cache Improvements
- 18:40Conclusion
related talks
33mRendering on the Web: Performance Implications of Application Architecture
29mCreate instant loading mobile web pages with AMP
35mBuilding lightning-fast sites for the mobile web
42mBuilding quick apps in 2018
46mSpeed Matters: Client Side Performance Improvement Tips & Tricks
32mGetting 'Up To Speed' with Accelerated Mobile Pages
26mImproving Performance in Matu
28mThe Browser Hackers Guide To Instantly Loading Everything