The resource closure operator is a new concept within autonomic computing that differs from other approaches in this field in that it is not based on prediction. Earlier models have an assumption that its knowldege about the system is comprehensive enough to be able to predict future behaviour. Instead, the resource closure model acknowledges that it is not possible or even necessary to have such knowledge to be able to manage the system well.
This masters thesis will make developments to a model based on the resource closure operator, and implement it in a specific case of autonomic power management. The chosen case is dynamic processor frequency scaling, which is a method for reducing the processor's power consumption. The ultimate hope of this research is to contribute knowledge towards the goal of better utilisation of computing resources, and, eventually, towards the goal of reducing the overall power consumption of computer systems and data centers.
Three additions to the model will be presented to further enhance it and make it suitable for the chosen application. A proof of concept implementation of frequency scaling will be performed to show the feasibility of such an approach. Estimates of potential energy savings indicate that the use of a resource closure model is a viable approach for autonomic power management.