A Case Study of Service-Oriented Software-Architectures

Astract:

A formal approach to service-oriented architectures enables the architects to build high-quality, provably correct systems by enforcing formal discipline on the inter-component scale. This thesis undertakes an evaluative critical-instance case-study of a music aggregation web service on the basis of prior research for an abstract model of software architectures based on components and services. The site Findwhatyoulike.com relies on its aggregation backend to allow its users to listen to a song at a music-streaming service of their choice. Therefore, the case study is an illustrative example for mashup services in general. Mashups are a new generation of web applications that improve information sharing by combining existing web resources and web application program interfaces (APIs). The primary goal of the study is to ensure provably correct metadata retrieval from music-streaming APIs in a resource-efficient manner. As the case study is supported by a corporation, the comprehensive definition of resource efficiency includes both the business as well as the IT perspective. In business terms, number of content provider integrations, development speed, task separation, and developer specialization are important objectives. As the value of aggregation is proportional to the number of content providers, the model shows how an abstract interface in a service-oriented architecture aids with the decoupling of content provider specific components. To formalize the desired decoupling the syntactic dependency relationship, the layered architecture style and, informally, graph theory is utilized. While minimizing the resources that run the web service is sought after from an IT infrastructure perspective, the more valuable resource for a mashup is network usage. In this regard, the thesis seeks a provably correct retrieval and storage process for a decoupled service architecture. Finally, for an academic researcher, the case study serves as a real-world example that aims to complement the developed abstract model and formal modeling techniques.