Roshni Malani’s interest in software engineering research began in the classroom. From this initial impetus, she pursued her project with Professor Richard Taylor and Eric Dashofy, one of his graduate students. Roshni’s experience with undergraduate research confirmed her desire to become a university professor. She is currently a student at University of California, San Diego working towards a doctorate degree in Computer Science. For those students who are considering an undergraduate research project, Roshni advises beginning their projects early and making them a priority. When she is not analyzing and interpreting data from engineering research, Roshni can be found playing tennis, skiing, and shopping.

Software architectural styles are the key to designing and implementing large software systems. Architectural styles impose rules and constraints on software architectures to elicit beneficial properties from those architectures. C2 is a component- and message-based style that promotes the development of loosely-coupled applications that are distributable, dynamic and adaptable. However, the constraints of this style are not well supported by modern object-oriented programming languages. The Java Message Service (JMS) standard, implemented by many middleware vendors, provides a number of capabilities needed by C2 applications. This study describes a novel software architecture framework for building C2 applications based on a JMS implementation. An empirical comparison of this framework with a hand-coded, non-middleware-based framework was performed. The results of this comparison demonstrated that although both frameworks exhibited a linear degradation in performance as the number of components was increased, the non-middleware-based framework performed better. However, the JMS-based framework has the potential to provide many other beneficial features.

Richard Taylor Donald Bren School of Information and Computer Sciences

An architectural style for significantly-sized software applications must be carefully chosen to ensure the final system will exhibit all the qualities the designers seek. Once chosen, appropriate implementation techniques must also be selected. It is commonly assumed that generic, commercially-produced infrastructure will be superior to open-source technologies. Roshni Malani’s project put this assumption to the test—and discovered that, in the domain she studied, the conventional wisdom was wrong. Commercial middleware can impose heavyweight constraints, leading to sub-par performance as compared to more limited, judicious designs. Projects such as this confirm again that the best approach a researcher can take is to question assumptions and follow the data!