Data access visualization for legacy application maintenance

Abstract

Software clustering techniques have been studied and applied to analyze and visualize the actual structure of legacy applications, which have used program information, e.g., dependencies, as input. However, business data also play an important role in a business system. Revealing which programs actually use data in the current system can give us a key insight when analyzing a long-lived complicated system. In this paper, we calculate indexes indicating how a data entity is used, making use of software clustering, which can be used to detect problematic or characteristic parts of the system. The developed technique can reveal the characteristics of a data entity; i.e., it is used like master data. We applied this technique to two business systems used for many years and found that our technique can help us understand the systems in terms of business data usage. Through case studies, we evaluated the validity of the indexes and showed that software visualization with the indexes can be used to investigate a system in an exploratory way.