Software Architecture Erosion and Performance Degradation in Banking Information Systems: Longitudinal Tracking

Abstract

The phenomenon of software architecture erosion presents a substantial challenge for large-scale enterprise systems, particularly within the highly constrained and mission-critical context of the banking and financial services sector. Over time, as systems undergo continuous maintenance, feature additions, and urgent bug fixes, the implemented architecture tends to diverge from its intended design, leading to a state of structural decay. While the theoretical consensus suggests that such architectural erosion negatively impacts system maintainability, there remains a critical gap in empirical research quantifying its direct, longitudinal effect on operational performance metrics such as response latency and resource consumption. This study addresses this gap by conducting a comprehensive longitudinal analysis of a core banking information system over a continuous sixty-month period. By leveraging automated static analysis to measure structural decay alongside dynamic application performance monitoring data, this research establishes a definitive empirical link between the accumulation of architectural smells and progressive performance degradation. The findings reveal that specific manifestations of erosion, most notably cyclic dependencies and the emergence of monolithic components, exhibit a strong positive correlation with increased transaction response times and elevated central processing unit utilization. The study provides crucial evidence that architectural debt is not merely a maintainability concern but a direct threat to system performance and operational reliability. These insights offer actionable intelligence for enterprise architects and technical leaders seeking to justify proactive refactoring initiatives and establish sustainable software evolution practices within complex, high-throughput environments.