Análisis comparativo de rendimiento entre fastapi y express.js bajo condiciones de red degradada

Abstract

This work evaluates the performance degradation and resource efficiency of FastAPI and Express.js under emulated network conditions. The study addresses the impact of telecommunications congestion on service reliability, a factor often overlooked. The work is aligned with the research line of Software, Information Technologies, and Data Science. To this end, a quantitative experimental design is employed in a hardware-isolated container environment. First, the baseline of both frameworks is determined under ideal conditions. Then, they are analyzed under latency and packet loss scenarios by injecting symmetric network chaos using statistical profiles from M-Lab Ecuador. Through transactional stress tests, efficiency is compared by measuring extreme latency, resource consumption, and garbage collection activity. The results show that a degraded network reduces the transactional throughput of both frameworks to equivalent levels, eliminating the advantage in terms of latency and responsiveness that FastAPI exhibits under stable conditions, even though the latter retains a significantly higher processing efficiency. Faced with connection retention, Express.js saturates the CPU due to the execution of garbage collection routines, whereas FastAPI demonstrates progressive memory retention. Neither framework alone is suitable for services deployed on degraded infrastructures; therefore, the implementation of fault tolerance patterns is required.