Continuous quality control of long lived software systems

Virtually any soware dependent organization has a vital interest in reducing its spending for sowaremaintenance activities. In addition to financial savings, formany organizations, the timeneeded to complete a soware maintenance task largely determines their ability to adapt their business processes to changing market situations or to implement innovative products and services. With the present yet increasing dependency on large scale soware systems, the ability to change existing soware in a timely and economical manner hence becomes critical for numerous enterprises of diverse branches. e ability of a soware system to be changed and extended in an efficient manner is commonly referred to with the termmaintainability. Despite its widely acknowledged importance, many soware developing organizations today do not explicitly define processes nor apply specialized techniques to ensure maintainability. is is especially precarious as the quality of soware systems typically undergoes a gradual decay in the process of their evolution and therefore needs to be controlled continuously. We claim that a major obstacle to a mature discipline of maintainability engineering is posed by the unsatisfyingly vague definitions of maintainability used today. Various approaches, usually in the form of quality models, have been proposed over the last four decades to remedy this problem. However, no comprehensive basis for controlling the maintainability of large soware systems in a continuous manner has been established so far. is thesis proposes a novel approach for modeling maintainability that explicitly associates system properties with the activities carried out during maintenance and thereby facilitates a structured decomposition of maintainability. e separation of activities and properties supports the identification of sound quality criteria and allows to reason about their interdependencies. As the activities are the main cost factor in soware maintenance, we consider this separation a crucial step towards the ultimate goal of a truly economically justified practice of maintainability engineering. e approach is based on a quality metamodel that supports a systematic construction of maintainability models and fosters preciseness as well as completeness. Furthermore, we describe how maintainability models defined by the presented metamodel can be operationalized in the maintenance processes to support continuous quality control. is includes the definition a maintainability assurance process based on the presented concepts and a set of supporting tools. ese tools enable the design of quality models based on our metamodel as well as the automatic generation of guideline documents to communicate quality requirements to the developers. To support quality assurance activities, the tools allow the generation of review checklists and provide an integration of a defined maintainability model with quality assessment tools used for automated analyses. We demonstrate the applicability of our maintainability modeling approach, the generation of developer guidelines and review checklists as well as the integrated use of quality assessment tools in multiple case studies carried out in academical and industrial contexts.