Dynamic Power Management for QoS-Aware Applications
M. Marzolla R. Mirandola
University of Bologna (Italy). Department of Computer Science.
Reducing the power requirement of large IT infrastructures is becoming a major concern. Energy savings can be achieved with hardware and/or oftware solutions; in particular, modern CPUs can operate at different power levels that can be selected by software: low power modes reduce energy consumption at the cost of lowering also the CPU processing rate. In this paper we address the problem of reducing energy consumption of a large-scale distributed application which must comply to Service Level Agreements. Specifically, we propose EASY (Energy Aware reconfiguration of software SYstems), an on-line algorithm for dynamically adjusting the processing speed of individual devices in order to minimize the energy requirement and at the same time allow the application to provide an average response time below a predefined threshold. Our algorithm uses a Queueing Networks performance model to drive the reconfiguration process, so that the number of individual reconfiguration actions is greatly reduced. We formulate the energy conservation problem as a Mixed Integer Programming problem, for which we propose a heuristic solution technique. Numerical experiments show that the heuristic produces almost optimal results at a substantially lower computational cost. Therefore, Energy Aware reconfiguration of software SYstems (EASY) can be effectively applied on-line to make a large system energy-proportional.