Automated Test Distribution Framework for Service Providing Systems

Abstract

Testing is still the only way to verify complex systems. As different technologies for testing small components of a system are established, e.g. unit tests, no common methodologies or approaches exists for testing the whole systems. Reasons for this purpose is the diverseness of the systems to be tested. But during development, especially the testing of whole systems, composed of distributed and communicating subsystems, is very important to find errors or weaknesses in the designs. Since the complexity of such systems causes an immense amount of testcases, an automated process is preferable. In this paper a general framework for automated system testing is introduced. It can be used for functional blackbox testing of any kind of systems which provide services that can be consumed by any kind of clients. Main issue is the automatical generation of an amount of service requests based on an abstract test-scenario. Since these requests has to be executed by the clients, a test-infrastructure is part of the framework and allows to communicate remotely with a dynamical number of distributed clients. Information about success of all requests are collected, analysed and can be used to influence the test-scenario. So it is possible, to provoke situations which indicates problems. The usage of this new framework allows to generate tests, which can help developers to find errors or weaknesses in the designed system. Furthermore it can be used to maintain or benchmark a system.