The Future of Efficiency Is Here: Add Planning to Your Schema!
summary
A new execution engine called Graphast builds declarative plan resolvers and composable step classes that plan and batch work ahead of execution, eliminating N+1 by design and letting steps dedupe and optimize themselves. Version 0.1 is MIT licensed and on npm with a resolver compatibility layer for gradual adoption; PostGraphile v5 is a rewrite on it and a demonstrated SQL for the same query is one quarter the size and runs in half the time. Plans are cacheable per document and are already running in production.
AI-written summary · about
chapters (17)
- 0:00Introduction to Graphast
- 1:26GraphQL inefficiency problem
- 2:26The N+1 problem and batching
- 4:28SQL efficiency comparison
- 5:38GraphQL execution model flaws
- 7:54How Graphast works
- 9:52Plan resolvers in TypeScript
- 11:57Step execution and batching
- 13:32Optimizing with lookahead
- 15:42Custom step lifecycle
- 16:57Advanced step optimization
- 18:30PostGraphile performance
- 20:24Graphast vs other techniques
- 21:45Adopting Graphast
- 22:49Project status and future
- 24:09Conclusion and sponsors
- 25:56Q&A session
related talks
23mBlueprints of Successful GraphQL Architectures
27mSupergraph Saving Lives
24mI’ve finally seen the light, the future is a graph
16mKeynote: GraphQL’s Next Chapter: Progress, Proposals, and Participation
25mGraphQL, the Domain Way
21mLightning Talk: Search and Execute with Code Mode Backed by the...
23mGradually Adopt GraphQL Without Writing any Backend Code
31mAccelerating GraphQL API Development with Schema Storming and Code-first Implementation