What makes one software architecture more testable than another?

Abstract

Software testing expends as much as 50% of development costs and comprises up to 50% of development time, yet most of the research in software architectures marginalizes the importance of testing in relation to architectural design decisions. It is our belief that testing has a dual role in software architectures, defect prevention and defect detection. How effectively testing fulfills its roles is dependent on three criteria, matching of software architecture to test strategy, the ability to detect certain fault types given a specific test strategy, and the interaction of software architecture, test strategies, and fault detection. We introduce what we believe to be the salient issues in achieving defect prevention and detection under the three separate criteria, and provide examples of interdependencies between software testing and software architectures. We conclude with suggestions for future work.