Software Engineering: The End of the Crisis?

Abstract

The chapter gives an overview of the concept of crisis and its application to software development and software product lifecycle. Managing software development lifecycle is a challenge for large-scale and mission-critical applications, especially in crisis. To solve the challenge, a lifecycle management methodology is required; the methodology includes models, methods and supporting tools. We discuss the general lifecycle pattern and its stages. We find that the cost of defect detection and fixing increases exponentially as we move from the earlier stages to the later ones, so error detection should happen as early as possible. Lifecycle models selection determines the mission-critical parameters of the project: the architecture of the project, its budget and timeframe. The model selection also determines product artifacts and quality attribute s; these are based on product quality metrics, which make crisis management more accurate and predictable. The lifecycle model selection should be adequate to the experience of the project team in terms of problem domain expertise and operational knowledge of technologies, tools and standards. We describe a number of lifecycle models, such as build-and-fix , waterfall , incremental, object-oriented and spiral. Some of these models require iterative development; the others are more straightforward. Certain models require a high level of discipline and organizational maturity. There is no universal model, which suits any software product equally well. The scope and size of the project are the determinants for lifecycle model selection; we can customize or combine certain models in order to adjust to the specific features of the project. For each model discussed, we identify the key advantages and disadvantages.