Requirements engineering: Foundation for software quality (REFSQ2020)

Abstract

Many organizations undertake large-scale projects of application migration due to availability of scalable and cost-efficient technologies. Such legacy application migration projects are very complex since the process involves in-depth profiling of the applications.During the initial profiling phase, it is imperative to understand the underlying complexities of individual applications, as well as the interdependencies among applications in the organization. This analysis phase can take considerable time and effort, depending on number and complexity of the applications. The main goal of this paper is to provide a framework that provides a cost-effective and quick approach to study the interdependencies between legacy applications with minimal prior knowledge of application usage.In this paper, we propose a framework that uses community detection algorithms and other established techniques from graph theory, to discover interdependencies of legacy applications within an organization, group these highly interdependent legacy applications in clusters, and finally sequence the clusters for migration to a modern platform. We study the proposed framework through three case studies, using network datasets from a large US organization.The experimental results from the proposed framework suggests that legacy applications can be grouped into clusters with high interdependencies between each other. Also, the framework shows how organizations can then appropriately sequence the clusters of legacy applications into a phase-wise migration project, thereby reducing migration costs.The proposed framework provides a valuable design input to organizations on how to determine the interdependencies between the various legacy applications that are in scope for migration to a modern platform. Such large-scale migration projects can be simplified and broken down to use a systematic approach, thereby reducing migration costs and data integrity challenges.