Abstract :
Because the complexity of today´s systems is so high, the changes in technology so fast, and the use of technology so broad, systems are no longer defined and bounded by their physical form, nor by their components. Instead, systems are defined by the characteristics of their use. Depending on the requirements, the quality of a system may vary from one configuration to another and from one usage to another. System testing, therefore, must be designed to include differences within and between application domains, and the testing process must be designed to evolve with changes in technology and requirements. This paper presents an approach to testing and evaluating large, complex systems and demonstrates it by using two real cases. The approach is based on usage models derived from the customer´s perspective. In contrast to traditional black-box and random testing, structure testing and controlled experiments are used to explore the non-functional aspects of systems
