Unlocking Performance with Response Caching
summary
Apollo rebuilt response caching inside the Apollo Router using Redis-backed router cache with TTL-based entries, per-subgraph and per-type controls, and a new cache debugger that shows freshness and sources. It adds cache tags for programmatic invalidation (HTTP curl API) so one user update can invalidate just that user’s entries, not the whole type. Demo shows subgraph latency dropping from ~250ms to ~100ms, production preview sees customers reduce pods, and the system handles ~85 billion requests/week.
AI-written summary · about
chapters (10)
- 0:00Response caching in Apollo Router
- 3:50Caching reduces latency and costs
- 7:24GraphQL vs CDN caching
- 10:36Current caching in Apollo Router
- 13:03New cache metrics and debugger
- 14:09Live demo: response caching
- 21:23Cache headers and TTL setup
- 24:33Programmatic cache invalidation
- 32:22Measuring performance improvements after caching
- 32:42Q&A
related talks
08mGetting started with GraphQL and caching on Next.js + Apollo + Layer0
21mCaching GraphQL APIs
37mCaching Deep Dive: The Ultimate Way To Speed up Your GraphQL API
24mGraphQL Clients
21mCaching your GraphQL data graph
23mHow to edge cache GraphQL APIs
24mModern Apollo Client React
08mAdvanced SSR Caching with React