Modeling and optimizing content-based publish/subscribe systems

Publish/subscribe systems are used increasingly often as a communication mechanism in loosely-coupled distributed systems. Their field of application is widespread ranging from communication platforms for real-time enterprises to Internet scale information dissemination systems. The performance of publish/subscribe systems plays an essential role for their acceptance. Especially, in mission-critical and financial areas, the system should guarantee certain predefined performance metrics such as throughput and latency. On the other hand, operators want to run publish/subscribe systems as efficiently as possible to reduce their operating costs. These requirements lead to the necessity to develop strategies for planning and managing publish/subscribe systems. In my PhD project I will introduce such strategies which target the complete life cycle of publish/subscribe systems including the design, the operation and the redesign phase. I will show how an optimized infrastructure meeting predefined requirements can be determined before the system is deployed and how this approach can be used for a redesign. This allows designers to decide about the required size and capabilities of the infrastructure they have to set up. Furthermore, I will investigate how the publish/subscribe system can continuously be optimized at runtime using a new kind of adaptive routing scheme. This way, the system will autonomously provide the best quality of service possible in dynamically changing environments and reduces management costs.