Architectural Modernization of Legacy PHP Systems : Performance, Maintainability, and the Role of AI-assisted Development

Abstract

Legacy applications built with a mix of HTML, PHP, and JavaScript are still widely used, but their tightly coupled structure makes them difficult to maintain, extend, and secure. This thesis explores how such systems can be effectively modernized by redesigning a representative part of the Digitized Commitment Management (DCM) tool into modular React and Node.js architecture. The goal was to improve maintainability and performance while introducing a clearer and more consistent application structure. The modernization effort included rebuilding the frontend as a component-based React application, introducing a dedicated Node.js API layer with parameterized SQL queries, and replacing ad-hoc PHP session handling with a Microsoft Entra ID authentication flow. These changes separated concerns that had previously been mixed, resulting in a cleaner, more reusable, and easier-to-maintain codebase. AI tools were used in a supportive role to speed up repetitive tasks such as scaffolding and pattern extraction. All generated output was manually reviewed to ensure correctness and security, making AI an accelerator rather than a decision-maker. The modernized solution was evaluated across performance, maintainability, security, and developer productivity. The results show clear improvements: significantly faster page rendering, reduced code duplication, stronger security through a centralized backend, and significantly more maintainable architecture. The results show that architectural redesign is the primary driver of modernization success. When combined with disciplined, review-driven AI assistance, modernization becomes both technically feasible and organizationally sustainable. The study demonstrates how legacy PHP systems can evolve into modular, maintainable, and secure platforms aligned with contemporary enterprise standards.