This research project focuses on the intersection of energy management and virtual machine environments. Contemporary energy management algorithms are implemented in operating systems and rely on detailed application, system, and device monitoring to manage devices and improve energy efficiency. Virtual machines make energy management more complicated in several ways. First, virtual machines are lacking application context. Second, guest operating systems lack global information. Third, virtual machines typically present virtualized devices to the guest operating systems, so that a guest operating system no longer has detailed device energy profiles. This project investigates the proper distribution of functionality between the guest operating system, virtual machine, and underlying virtual machine monitor so as to improve overall system energy efficiency. All guest operating systems must collaborate with the virtual machine monitor to exchange information about current context and the device state. System-wide energy efficiency requires global information, as is available in non-virtualized systems. The challenge is to provide the desired information flow while preserving the abstraction provided by the virtualization. Solving this challenge is critical since future systems will rely heavily both on energy efficiency to prolong battery life and reduce energy costs, and on virtualization to provide isolation, portability, and many other benefits. The tools and techniques developed in this project will provide a platform for future research. Furthermore, the goal of this project is to involve students in designing and evaluating energy-efficient virtual machines for various devices through in class assignments as well as independent studies. The projects and assignments will provide students with better understanding of the challenges that they will face in their future careers.
