Current Research Topics
The KBS group mainly deals with resource management problems in operating systems and distributed systems. While during the nineties the focus was on parallel computing, we are now working in the area of Grid-Computing. We are aiming at creating a comfortable and transparent infrastructure for Grid applications by mainly relying on self-organizing. That includes economic and nature inspired approaches. In particular, we are addressing scheduling of complex Grid applications, dynamic placement of logical resources, and self-adaptive information dissemination (Publish/Subscribe). The fed-back, reactive approach requires collecting metadata about the behaviour of the programs to make good dynamic resource allocation decisions at run-time. Furthermore, we provide metadata in form of stochastic models for the execution behaviour of Grid applications as a basis for predictive and proactive resource allocations. We also apply the method of Model Driven Architecture (MDA) not only to automatically generate code, but also to derive information about the program behaviour from the model that can be used for allocation decisions at run-time. We regard quality of service parameters (QoS) as timeliness, dependability or security as essential for Grid environments and explicitly consider QoS aspects in our research. In addition, we are interested in composability of QoS, i.e. how QoS properties of a system can be derived from QoS properties of its components.