Applicability of Software Reliability Growth Modeling in the Quality Assurance Phase of a Large Business Software Vendor

Software reliability growth modeling aims to use data on experienced failures for prediction of future quality levels. We analyze the applicability of this approach in the context of business software that is still in the quality assurance phase and has not been released to the customer yet. We find that the approach is not applicable in our research setup that is characterized by an agile process organization. We identify qualitative reasons why the models' assumptions are violated in our industrial case study and conduct a quantitative analysis whether data availability challenges can be addressed by mapping issue tracking data as the only available data source to the required system under test execution time. The derived metrics support managers in their decision whether their processes are suited for software reliability growth modeling.