Using Client-Side GraphQL Data to Reduce Server Costs

Sabrina WassermanMeta, Software Engineer

In this talk, I’ll discuss how we utilize an in-memory client-side GraphQL data store at Meta to reduce unnecessary network fetches, and reduce server latency for executed requests by dynamically skipping fields that can be fulfilled by our in-memory store. Specifically, I'll cover: - What are Abstract Syntax Trees (ASTs), and how do they allow us to determine if a query's data is fulfillable from an in-memory data store. - GraphQL Partial Rendering - How we at Meta are able to immediately vend partial query responses from an in-memory store for more response UIs. - How we dynamically skip client-side fulfillable fields, and only fetch missing fields from the server to reduce server costs