Service virtualization using recorded interactions

Abstract

Today’s enterprise software systems are much more complicated than the past. Increasing number of dependent applications and heterogeneous technologies makes testing of such systems challenging due to multiple reasons including unavailability of components, high cost of using services and conflicting schedules of different develop ment teams. Therefore in such software systems, it may be more convenient to use virtual components instead of the real ones. Service virtualization is a technique to mimic the behavior of a real component. Services are classified into two groups namely; stateful and stateless services. In this thesis, we introduce techniques for creating vir tual copies of both stateful and stateless services. To the best of our knowledge, this is the first work to create virtual services for stateful services. We employ bioinfor matics and machine learning algorithms in developing our solutions. We demonstrate the validity of our approaches on data sets collected from real life services and obtain promising results.