Energy Efficient Management and Scheduling of Computational Resources

Abstract

Information Technology has a big impact on climate change and global warming as the current and future challenges of the world. In this PhD thesis, we focus on green computing to address these challenges. With the advent of new technologies, we need to explore these technologies to see if they can address green computing. The first stages of this thesis uncover the flux of technologies and its impact on green landscape. We envision the duality of green computing with technological trends in other fields of computing such as high performance computing (HPC) and cloud computing from one hand; and economy, and business, on the other hand. In addition, IT’s energy consumption and sustainability impact are expected to increase. Contemporary technologies are moving towards Intelligent Computation in order to optimize resource and energy consumption. Intelligent Computations are done with the techniques and mechanisms of new computing technologies such as infrastructure-adapted-to-applications, virtual machine consolidation, optimization algorithms, hardware and software co-design, and application profiling to optimize resource consumption, and pay-as-you-go business model to reduce costs, etc. In green world, research on minimizing energy and resource consumption through algorithmic and software techniques such as monitoring, power-aware consolidation, scheduling, optimization algorithms such as bin packing as well as user/application profiling and debugging are other aspects of addressing energy efficiency. These facets of Intelligent Computation constitute the main parts of resource management. In the second part of this Thesis, we address some aspects of Intelligent Computation as part of resource management and scheduling. More specifically, we introduce the problem of resource management more precisely and describe computing system problems from the resource management point of view. We explore resource management components. Then, we model resource contention metric that is one of the main metrics explored in this thesis. We develop effective energy aware consolidation policies. Finally, we propose a novel autonomic energy efficient resource management and scheduling algorithm