Autonomic Energy-Aware Tasks Scheduling

Abstract

The increasing processing capability of data-centers increases considerably their energy consumption which leads to important losses for companies. Energy-aware task scheduling is a new challenge to optimize the use of the computation power provided by multiple resources. In the context of Cloud resources usage depends on users requests which are generally unpredictable. Autonomic computing paradigm provides systems with self-managing capabilities helping to react to unstable situation. This article proposes an autonomic approach to provide energy-aware scheduling tasks. The generic autonomic computing framework FrameSelf coupled with the CloudSim energy-aware simulator is presented. The proposed solution enables to detect critical schedule situations and simulate new placements for tasks on DVFS enabled hosts in order to improve the global energy efficiency.