School of Computing Science


DOPCHE - Dynamic Operating Policies for Commercial Hosting Environments

Amongst the areas of research vital to the development of infrastructure support for eScience is the provision of systems with predictable, differentiated levels of performance and dependability. It is essential that systems be able to provide sustainable levels of service whilst adapting their operating policies to dynamic workloads, resource pools and system configurations. While policy making, prediction and monitoring, and self-tuning infrastructures are available, they exist to different degrees and to different extents, and there is no established framework in which they are (or can be) currently integrated. In future systems it will be imperative that these are provided as fundamental, integrated components of a combined infrastructure.

This project addresses research in the areas of: (1) systems and application modelling that allow us to predict reliably the behaviour of future eScience infrastructures; (2) performance verification that provides evidence-based validation on the delivery of these services; (3) self tuning and adaptive systems that aim to reduce the cost and complexity of managing the infrastructure in the light of changing user needs and changing infrastructure support.