An Innovative Metric-based Clustering Approach for Increased Scalability and Dependency Elimination in Monolithic Legacy Systems

Scalability is one of the system's characteristics highlighted in the recent literature, and it is directly related to issues that are encountered in state-of-the-practice technology. The scalability of a system is challenging because monolithic legacy systems are hard to scale due to the high level of component dependencies. To the best of our knowledge, there is no published work available that can identify the components from a monolithic legacy system in the context of dependent and independent components and scale them accordingly. The main contribution of this paper is the proposal of a novel approach for the exclusive identification of dependent and independent monolithic legacy system components. The proposed approach also helps to remove the dependency among components of monolithic legacy systems. As a result, it establishes a precise method that identifies all the components of an application and removes the dependency among components, helping to increase the scalability of the resulting application. This approach was validated by several experiments, and the key findings were the identification of dependent and independent components, the identification of relationships among components, and the identification of the abstract level architecture of the monolithic legacy system. In future work, the proposed method will be enhanced toward the recovery of the whole system's architecture.