Bachelor's thesis (paper and presentation) about the development of algorithms for the dynamic right-sizing process in data centers. PDF: https://github.com/kappelmann/bachelor_thesis_algorithm_dynamic_right_sizing_data_centers/blob/master/thesis/build/main.pdf
A significant fraction of modern data centers' energy costs is due to their high energy consumption during periods of low load. One possible solution to deal with this waste of energy is to dynamically adjust the number of active servers and to efficiently distribute the varying workloads, that is to dynamically right-size the data center. In this thesis, we derive algorithms capable of finding approximative as well as optimal solutions for this dynamic right-sizing process by reducing the data centers' scheduling problem to the shortest path problem of acyclic graphs. All developed algorithms are proved for correctness and made accessible by using pseudo-code that can be easily transferred to arbitrary programming languages.
Kevin Kappelmann
E-mail: kevin@kappelmann.me
Website: https://kappelmann.me