Connectors Deployment Overview

This overview walks through end-to-end deployment workflow for Apollo Connectors. First, it's helpful to understand Connectors' architectural approach and what this means for deployment.

Connectors architectural approach

Unlike traditional GraphQL implementations, Connectors don't require you to deploy GraphQL servers. Instead, you manage:

• GraphQL schema files: Define your API and publish to GraphOS (stored in your source control)

• GraphOS Router configuration: YAML files that configure your graph's runtime (also in source control)

• Router runtime: The executable binary that you deploy to your infrastructure

The GraphOS Router

The GraphOS Router serves as the runtime environment that powers your Connectors implementation. It's a high-performance service that:

• Receives GraphQL operations from your clients

• Creates and executes a query plan to retrieve data from your backend services

• Assembles responses from multiple sources into a single response for the client

• Handles authentication, caching, and other cross-cutting concerns

Since Connectors eliminate the need for passthrough GraphQL servers, deployment focuses solely on hosting and running the router.

Deployment workflow

The deployment workflow for a graph with Apollo Connectors involves:

1. Local development - Define your GraphQL schema using Connectors

2. Router configuration - Configure the GraphOS Router for your security, performance, and observability needs; you can also include variables in your configuration that you use in your schema when configuring requests.

3. Schema publication and management - Publish schema updates to GraphOS

4. Router deployment - Deploy the GraphOS Router to your infrastructure

Local development

During local development, you:

• Create GraphQL schema files with Connectors directives

• Test queries locally using the router's dev mode

• Iterate on your schema design

The Rover CLI provides the dev command to validate your schema and run a local development server.

Router configuration

The GraphOS Router uses a declarative YAML configuration model that provides control over:

supergraph:
  listen: 0.0.0.0:4000
headers:
  all:
    request:
      - insert:
          name: "x-api-key"
          value: "${env.API_KEY}"

Key configuration areas include:

• Security, including authentication, TLS settings, and request limits

• Performance, including traffic management and shaping, caching strategies, and batching and deduplication

• Observability, including logging configuration, telemetry, and distributed tracing

All router configurations are supported for graphs using Connectors, except for those documented in the limitations.

Schema publication

Publishing your schema to GraphOS involves:

• Using Rover to publish schema changes

• Integrating schema checks into CI/CD pipelines

See the Schema Publishing overview for more information.

Router deployment

The GraphOS Router is a lightweight, Rust-based service that can be deployed alongside your existing infrastructure:

• Docker - Run the router as a containerized application

• Kubernetes - Deploy using Kubernetes manifests or Helm charts

• Cloud services:

  • AWS - Deploy on ECS or other AWS services

  • Azure - Run as an Azure Container App

  • GCP - Deploy with Google Cloud Run

Deployment best practices

Best practices for this workflow include:

• Version control - Store your schema and router configuration in source control

• Environment management - Use separate graph variants for each environment and configure environment-specific values

Environment management

Each variant has its own schemas, along with its own change history and metrics.

You can use environment variables to manage different values per environment:

1supergraph:
2  listen: 0.0.0.0:${env.PORT:-4000}
3
4headers:
5  all:
6    request:
7      - insert:
8          name: "x-api-key"
9          value: "${env.API_KEY}"  # Different per environment
10
11connectors:
12  sources:
13    products:
14      url: "${env.PRODUCTS_API_URL}"  # Different URL per environment

Next steps

• Learn more about how you can use router configurations to override request URLs and headers

• Refer to environment best practices for more detailed guidance on environments

• Explore router deployment options for your infrastructure

• Set up monitoring and observability for your deployed Connectors