Performance

The Performance pillar focuses on improving your supergraph's runtime performance. A well-architected supergraph platform not only delivers a great end-user experience. It also provides ways for its developers to proactively improve both real and perceived performance. Additionally, it gives developers the documentation and tools they need to ship and measure those changes effectively.

Principles

Enable developers to improve perceived end-user latency

Improving perceived end-user latency is crucial for organizations that rely on digital services. Slow loading times can cause high bounce rates and lower conversion rates, hurting revenue. These are some techniques developers can use to improve perceived end-user latency:

• Using response compression
• Prioritizing data with @defer
• Using subscription/real-time GraphQL features

Configuring trace sampling and using persisted queries can further also help reduce request sizes and latency. Field-level latency metrics can also help graph consumers make informed choices about operation shapes.

Enable subgraph developers to improve graph performance

Subgraph developers can also help improve the performance of the supergraph by optimizing their subgraph's performance. Best practices include:

• Leveraging caching to improve the performance of operations and resolvers
• Using dataloaders in resolvers
• Implementing Automatic Persisted Queries (APQs) to improve subgraph request latency

Find opportunities to improve performance

Observability tools can be invaluable in identifying opportunities to improve the supergraph's performance. Resolver execution metrics and distributed traces aid in optimizing the performance of individual components. Additionally, operation traces can help identify bottlenecks and inefficiencies across the system. By leveraging these observability tools, teams can proactively address performance issues before they impact end-users.