Bridging the application knowledge gap

Abstract

Regarding energy efficiency, resource management in complex hard-and software systems that is based on the information typically available to the OS alone does not yield best results. Nevertheless, general-purpose resource management should stay independent of application-specific information.To resolve this dilemma, we propose a generic, ontologybased approach to resource scheduling that is context-aware and takes information of running applications into account. The central task here is to recognize situations that might necessitate an adaptation of resource scheduling. This task is performed by logical reasoning over OWL ontologies. Our initial study shows that current OWL 2 EL reasoner systems can perform recognition of exemplary situations relevant to resource management within 4 seconds.