I had the privilege of speaking at QCon 2025 in the Microservices Track. My topic was “High-Performance API Design in Practice.” Here are the key takeaways from my talk and the audience discussions.
Takeaway 1: API Design is System Design
Many engineers treat API design as a surface-level concern. The reality is that API design decisions affect database schema, caching strategy, load balancing, and even team structure.
Takeaway 2: Performance Starts at the Contract
Design the API contract with performance in mind. Use pagination for list endpoints. Use filters to reduce payload size. Use sparse fieldsets to let clients request only what they need.
Takeaway 3: Implement Caching at Multiple Layers
CDN for static assets and read-heavy endpoints.
Redis for session data and database query results.
Application cache for repeated computations.
Do not rely on a single caching layer.
Takeaway 4: Use Async Patterns When Appropriate
Not all endpoints need to be synchronous. For heavy operations, return a 202 Accepted with a status URL and let the client poll for completion.
Takeaway 5: Test Under Realistic Load
Use tools like k6 or JMeter to simulate production traffic. Test with realistic data volumes, not synthetic datasets that are too clean.
Audience Questions
The most common question was: “How do you balance performance with maintainability?”
My answer: Performance is a feature, not a trade-off. Good design achieves both.