Apollo Federation入門

Welcome to Apollo Federation! This powerful architecture helps you build a single, unified GraphQL API from multiple independent microservices.

Imagine having many small services, each responsible for a part of your application. Federation lets them all contribute to one big GraphQL schema.

In a microservices architecture, you often have different teams managing separate backend services. How do you expose a coherent GraphQL API?

• Monolithic GraphQL: One big server, hard to scale for large teams.
• Multiple GraphQL Endpoints: Clients need to query different services, increasing complexity.

Federation offers a clean solution to this challenge.

The core idea of Federation is that different services (called subgraphs) can each define a part of your overall GraphQL schema.

These parts are then combined by a central Gateway into one unified graph. It's like having many contributors to a single, shared blueprint.

Apollo Federation involves two main architectural components:

• Gateway: This is the entry point for all client requests. It knows about all your subgraphs and how to route queries.
• Subgraphs: These are your individual microservices. Each subgraph implements a specific part of the GraphQL schema and resolves data for its domain.

When a client sends a GraphQL query to the Gateway, here's a simplified flow:

1. The Gateway receives the query.
2. It analyzes which subgraphs are needed to fulfill the request.
3. The Gateway breaks down the query and sends specific parts to the relevant subgraphs.
4. It then combines the results from all subgraphs into a single response for the client.

Federation offers significant advantages for large, evolving systems:

• Scalability: Teams can develop and deploy subgraphs independently.
• Decoupling: Services remain autonomous, reducing inter-service dependencies.
• Unified API: Clients get a single, consistent GraphQL endpoint.
• Strong Type Safety: The entire graph is strongly typed, even across services.

You might have heard of Schema Stitching, another way to combine GraphQL APIs. How does Federation differ?

With Stitching, individual schemas are combined at the Gateway, but the sub-schemas are often unaware of each other. It's more like joining separate documents.

Federation promotes a truly distributed graph where subgraphs explicitly collaborate and extend shared types.

The execution model is another key difference:

• Stitching: The Gateway might fetch data from one service, then another, and manually combine them. This can lead to less optimized data fetching.
• Federation: The Gateway understands the relationships between types across subgraphs. It can intelligently plan efficient queries, fetching only what's needed from each service.

Federation is particularly well-suited for:

• Large organizations with many development teams.
• Complex microservice architectures that need a unified API.
• Scenarios where you want to expose a single, evolving GraphQL graph over time.

If you have a smaller, simpler application, a single GraphQL server might suffice.

When building federated subgraphs with Spring Boot, you'll define your GraphQL types as usual. However, you'll also use specific Federation directives.

The most important is @key, which identifies an entity's primary key across different subgraphs. This allows the Gateway to understand and resolve relationships.

Let's check your grasp of the core concepts.

Great job! In this lesson, we introduced Apollo Federation, a powerful approach for building a unified GraphQL API from multiple microservices.

• We covered the roles of the Gateway and Subgraphs.
• We explored how Federation enables a distributed schema and efficient query execution.
• We also saw how it differs from Schema Stitching.

Next, we'll dive into building your first federated subgraphs with Spring Boot!