graphql-architect

Federated GraphQL schemas use Apollo Federation to split your API across subgraphs, each owned by a team. Define entity types with @key directives, establish clear boundaries between services, and use a gateway to compose and route queries. This enables independent service development while maintaining a unified graph.

Query optimization combines DataLoader for batch loading, complexity analysis to prevent expensive queries, and subscription scalability patterns. DataLoader deduplicates and batches resolver calls, complexity analysis sets execution limits, and subscriptions use efficient pub/sub to handle real-time updates across federated services.

A federation gateway composes subgraphs into a single schema using query planning and entity resolution. Configure composition rules, map entities across services via @key directives, set up query routing, and enable introspection. The gateway handles cross-service field resolution and error handling transparently to clients.

Yes, subscriptions work in federated architectures by routing subscription requests through the gateway to the appropriate subgraph. Implement subscriptions at the service level, configure the gateway's subscription transport, and use domain-driven patterns to ensure scalable pub/sub across teams and services.

Schema-first development defines the GraphQL schema explicitly, then generates code and types—enabling design review and federation contracts upfront. This approach supports type safety, automated code generation, and clear subgraph boundaries essential for federated multi-service APIs.

DataLoader batches and caches resolver calls to eliminate N+1 queries. Register DataLoaders for each entity type, group database requests from multiple resolvers, and execute them in a single batch. This is critical for performance in federated setups where subgraphs may chain entity resolutions.