Published Paper Details:

Microservices architecture has become the dominant paradigm for building scalable, independently deployable cloud-native applications, and orchestration platforms such as Kubernetes have emerged as the de facto standard for managing the lifecycle of these distributed services. As the number of microservices grows, an API gateway becomes essential for consolidating cross-cutting concerns such as routing, authentication, rate limiting, load balancing, and observability at the edge of the system. However, the proliferation of API gateway solutions -- including Kong, NGINX, Traefik, Spring Cloud Gateway, and the Istio Ingress Gateway -- makes the selection of an appropriate gateway for a given orchestrated deployment a non-trivial engineering decision. This paper presents an empirical evaluation of five widely used API gateways deployed on a Kubernetes-orchestrated microservices testbed. We measure and compare latency, throughput, error rate, CPU utilization, and memory footprint under varying levels of concurrent load, and we qualitatively compare the gateways on feature dimensions including service discovery integration, protocol support, extensibility, and operational complexity. Results indicate that NGINX-based gateways deliver the lowest latency and highest raw throughput for simple routing, Kong offers the strongest balance of performance and plugin-based extensibility, Traefik provides the most seamless Kubernetes-native service discovery, while Spring Cloud Gateway and the Istio Ingress Gateway trade raw performance for deeper integration with application-level and service-mesh-level concerns respectively. The paper concludes with guidance for architects choosing a gateway based on workload characteristics, team expertise, and operational priorities.