Guideline-based approach for improving the achievement of non-functional attributes of software

Abstract

As a response to business forces pushing for more quality, the aim of software engineering has gradually changed to develop functional software systems with high quality. One aspect of software quality is the non-functional attributes of software such as availability, usability, performance and maintainability. Software systems are usually required to possess multiple non-functional attributes in a certain priority order, with a certain level of quality. For achieving a non-functional attribute, some software engineering guidelines (such as architectural styles and design patterns) can be applied while developing a software system. However, finding the suitable set of guidelines to apply for achieving more than one attribute is not a straightforward job, due to the different effects that guidelines have on the quality of these attributes, as well as the relationships among the guidelines themselves. Inappropriate use of guidelines may result in software defects associated with these attributes. This dissertation presents a guideline-based approach for improving the achievement of non-functional attributes by avoiding defects due to incorrect application of guidelines. The approach guides the software engineer throughout all software development phases by providing a suitable set of guidelines to be applied. The set of guidelines provided conform to two main conditions: they form a homogeneous set without any overlapping of conflicting relationships, and they support the achievement of the required non-functional attributes proportionally to the attributes’ priorities. Applying this set of guidelines reduces defects associated with non-functional attributes by avoiding overlapping or conflicting guidelines, which results in better achievement of these attributes and higher-quality software systems. The approach is validated through both expert and user reviews as well as an example of a fictitious case study. Its output is verified using mathematical theory.