Stochastic modelling for energy efficient supercomputing

This thesis studies mathematical stochastic modelling frameworks to improve the energy efficiency in cluster computing systems, which are facing the challenge of growing energy consumption due to the fast-increased computing demand from phenomenon such as Big Data or the Internet of Things. These frameworks are developed with the reference to the analysis of empirical data collected from real computing systems. This work is a contribution to the engineering effort to make the modern Information and Communication Technology infrastructure more sustainable.