Abstract:
When building distributed systems, choosing the right API gateway architecture becomes critical for performance, security, and operational simplicity. In this session, we focus on the several patterns used to achieve this goal, focusing on Edge gateways (Nginx, Envoy, Traefik, etc). We’ll look at what they bring to the table, benchmark request latency, throughput, and resilience under load using a simple load tester with a Prometheus monitoring setup. We would also look into the sidecar architecture and when to use each pattern.]

Key Takeaways:
- Performance trade-offs: Measured latency & resource overhead for edge vs. sidecar.
- Security patterns: Implementing TLS termination, mutual-TLS,
and zero-trust in each model.
- Operational impacts: Simplifying policy changes, versioning, and scaling.
- Deployment recipes: Step-by-step for Envoy at the edge, Istio sidecars, and CI/CD integration.

The session concludes with integrating this architecture with Kubernetes. Then we have a quick look at use cases: Uber, Netflix, and Amazon’s use of API gateways to serve their customers.