Graphql. Accelerated
summary
Graphs suffer CPU and architecture costs, not just convenience: profiling with clinic flame shows most time goes to parsing/validation of the large query AST and repeated traversal. A pragmatic win is to cache parsed ASTs and validation results for hot queries and to compile frequent queries to optimized execution functions (using existing JIT libraries) — that combination yields multi‑fold throughput improvements over naive GraphQL servers while preserving schema typing and tooling.
AI-written summary · about
chapters (11)
- 0:00Start
- 1:34GET/facebook-friends-names?
- 1:57Nonstandard
- 2:39GET /facebook-friends-and-likes?
- 4:57GraphQL has a rich ecosystem
- 7:34Makes prototyping easy!
- 7:56GraphQL comes at a cost
- 13:07Uses an LRU to cache stable queries
- 16:51We did additional performance analysis using clinic flame
- 17:28On the fly code generation
- 18:34Using our improvements GraphQL becomes very competitive to rolling your own stack
