Agile Software Development of Information Systems in Support of Object Databases

Agile development practices have become widely adopted in the past decade. They are increasingly used in industry and have been subject to a wide range of research. The promises of agile development practices range from customer satisfaction and adaptivity to changing requirements to developer motivation and reliable software through iterative development and customer involvement. However, what has received little attention so far is the effect of agile development practices on the development of information systems where the static nature of databases conflicts with the agility of the development approach. One feature of agile projects is the iterative development approach by which users may start using the software as soon as minimal functionality is available. In the case of information systems, this results in the creation of databases with possibly valuable data. Together with the agile principle of frequent release of software updates, this leads to a need for frequent database evolution of customer databases which results in increased development effort. At the same time, the tight coupling of the incremental application development with application usage gives rise to new opportunities for exploiting the agile principles in favour of an improved development process. In this thesis, we explore various effects of agile development practices on the development of information systems and investigate new ways in which information system development can benefit from agile practices. We start with a case study of a large software development project of the European Space Agency (ESA) in which the author was involved. The project developed an information system for operating the ESA Herschel Space Observatory and processing observation data. The information system was developed using an agile approach involving development teams located in several different countries and research institutions. The development approach of the project was such that users used the system from early on in the development period while creating significant amounts of data and databases that needed to be continuously evolved in order to be compatible with the evolving data model and to be available later when the software became operational. This led to roughly bi-monthly database evolution being necessary for hundreds of databases throughout six years of development. Based on this case study, we highlight several issues and concepts related to the agile development of information systems and explore solutions and ways to benefit from agile development practices. For example, we discuss the suitability of object databases for agile development due to their approach of avoiding separate logical and conceptual data models with their associated mappings, thereby reducing the effort of evolving the data model. This tight coupling of application data model and database also has the advantage that it allows the tracking of navigation between objects and classes. This adds a dimension to profiling data that permits profiling on the level of the conceptual model and which is more difficult to obtain from relational databases. We visualise this profiling data