Locating Software Features in a SOA Composite Application

This paper describes the use of Feature Sequence Viewer (FSV) to perform feature location in a teaching and research program suite named Open SOALab. In this context, a software feature refers to software components that provide specific functionality. The composite application encompasses a system in which hotel brokers identify rooms meeting various criteria from among several hotel chains in multiple countries, and then exchanges the necessary amount of currency, using a currency broker to get several quotes and select the best one. The currency broker in turn uses two services: an authentication service and a settlement house. The various service interfaces are exposed via WSDLs. The system, running on Apache with php and nuSOAP, uses Apache's forensic log module and microsecond time stamps to generate data that is input into FSV which produces a browsable graphical representation of the messages in the system. FSV employs a component relevance index (pc) that is used to determine which messages are displayed within the viewer. A value of pc is computed for each message in the data set loaded into FSV. A user can raise and lower the threshold value for pc so that messages with pc values above the threshold are displayed and those below are not displayed. Three experiments of increasing complexity were performed to demonstrate the ability of this approach to extract feature messages by separating them from irrelevant messages within a SOA composite application.