How to Think like a Performance Engineer
summary
Starts with concrete tooling and numbers: use the Chrome User Experience Report and Treyo to learn that real users hit ~3.2s P75 where Lighthouse sometimes reports 21s, and that WebPageTest scripting finds hidden costs like a 1.4s redirect or a 3.8s DOMContentLoaded. Run lab tests in WebPageTest, tune connection/device to match P75/P95, and script realistic journeys (accept cookies, prime localStorage, click soft navs) not just cold URL loads.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:47Talk title meaning
- 4:39Key tools
- 9:14Agreeing on metrics
- 12:01Why time to first bite is a critical metric
- 14:07DOM content loaded
- 17:5875th percentile too low
- 19:21Reliable repeatable tests
- 22:28Lab tests too pessimistic
- 27:44Cold start problems
- 31:11DevTools application panel
- 33:43Scripting user interactions
- 38:40Cache misses on writes
- 40:47SPA soft navigations
- 43:36Test design takeaways
- 44:19Q&A
related talks
18mWebsite performance from user perspective
29mPerfecting Perf: Rapid, Rigorous and Reproducible Performance Experiments
48mThink Fast, First: Empowering a Performance Culture
35mWeb Application Performance Testing from Users Perspective
57mSite-Speed That Sticks
55mI Feel the Need for Web Page Speed
21mThe Future of Performance Tooling
16mFrom beginner to expert in web performance