Impact of Architecture and Quality Investment in Software Product Line Development

Abstract

Investment in architecture and quality improvement for a software product line can increase reuse, and consequently reduce effort, enhance product reliability, and shorten time- to-market. Such investments should be carefully chosen to be effective, to avoid over-investment, and to return benefits within the desired time. In this paper, we show how a stochastic simulation model can be used to explore the impacts of such investments. The model is validated by comparison to COPLIMO, a COCOMO II based effort estimation model for product line development, and by inspecting effort distributions of the generated unplanned work. For the illustrative model and scenarios in this paper, we show that the degree of architecture reuse has the largest impact. Preventing degraded architectural dependencies itself does not have a meaningful impact, but if such degradation is also associated with adverse effects on defect injection and detection, it can be significant. Process improvement has a meaningful impact, but over-investment is possible.