Uma análise comparativa do desempenho de frameworks back-end para APIs RESTful em ambientes controlados

Abstract

The increasing demand for scalable web applications has driven the adoption of different backend frameworks for the development of RESTful APIs. In this context, the choice of framework can directly impact application performance, stability, and resource consumption. This work presents a comparative performance analysis of REST APIs implemented using Node.js with Express.js, Django REST Framework, and ASP.NET Core under controlled load conditions. Three functionally equivalent REST APIs were developed based on a room reservation system, ensuring identical business rules, data structures, and CRUD operations. The applications were subjected to stress, spike, and endurance tests using Apache JMeter to simulate concurrent users, while Prometheus and Grafana were employed to monitor CPU and memory usage. All experiments were conducted in a containerized environment using Docker, ensuring reproducibility and standardized execution conditions. The results revealed significant performance differences among the evaluated frameworks. ASP.NET Core consistently achieved the best overall performance across all test scenarios, exhibiting higher throughput and lower response times. In the stress test, ASP.NET Core processed up to 280.4% more requests per second than Node.js and 1137.7% more than Django REST Framework, while reducing the average response time by up to 94.3%. Node.js demonstrated intermediate performance, standing out for its efficient resource utilization, whereas Django REST Framework showed considerable limitations under high concurrency, with higher resource consumption and reduced stability.These findings indicate that ASP.NET Core is more suitable for high-concurrency RESTful applications, while Node.js represents a balanced alternative in environments with resource constraints. Django REST Framework is better suited for scenarios with lower concurrency requirements.