Developing Models For The Purpose Of Understanding Legacy Systems

Abstract: In the context of this thesis, legacy systems refer to old computer programs that are highlyvaluable to a company. Maintaining these legacy systems is costly and can consume up to 80%of companies’ budgets [1]. Understanding legacy systems is essential when maintaining them,which consumes large parts of companies’ budgets. Facilitating legacy system understandingcan reduce the cost for IT companies. Previous research on legacy system understanding uses models, such as UML models. Suchmodels can not be interacted with by the developer. Instead, this thesis uses models that aresimplified executable computer systems that can be interacted with. This thesis aims atevaluating if such models can be used to aid developers in understanding legacy systems. Amodel was created based on a web-based game using reverse engineering. Participants withprogramming knowledge were interviewed about their experience of understanding the legacysystem versus understanding the legacy system after interacting with the model. The data fromthe interviews were analyzed using the reflexive thematic analysis method described by Braunand Clarke [2]. The results show that experienced developers were aided in understanding asoftware system after interacting with a model of the system. Therefore, it appears to be a goodidea to use models when trying to understand more complicated systems such as legacysystems. However, before this technique could be presented to companies in the IT industry,more research would have to be done into the amount of money saved compared to the costs ofcreating the models