'Divide et impera' with GraphQL and SSR
summary
A migration case study of splitting a half‑million‑line Python monolith into a Node.js server‑side rendering layer and componentized React UI, using GraphQL with Apollo to request exactly the needed fields. The team built a custom SSR renderer plus GraphQL batching and N+1 query fixes, measured ~5 ms average render per component under peaks (≈100k companies, ≈80M products), handled ~5k req/s and ~200 server renders/s (Googlebot spikes to 600–700), diagnosed memory leaks with ~300MB limits causing restarts, and observed lower bounce rates and longer sessions after incremental rollout.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 7:24Monolith problems
- 12:44GraphQL with Apollo
- 19:49React Router rendering
- 27:24SSR tips
- 32:03Migration issues
- 35:14Migration benefits
related talks
30mReact state management in a GraphQL era
31mLessons from Scaling GraphQL to Half a Billion Requests per Minute
22m5 Antipatterns that Made Our React/GraphQL App Slow (And How We Fixed Them)
22mReact State Management in a GraphQL Era
58mCrafting compelling real-time web experiences with GraphQL and React
28mGraphQLifying React Native App
24mReact State Management In a GraphQL Era
27mReact State Management In a GraphQL Era