ERES: An Energy-Aware Real-Time Elastic Scheduling Algorithm in Clouds

Abstract

Clouds have emerged as an innovative paradigm that aims to provide reliable, customized and elastic computing resources for users in an on-demand manner over networks. Meanwhile, green computing has become a major concern for the Cloud designers and administrators. However, previous energy aware scheduling algorithms in Clouds mainly emphasize on the energy saving based on the system workload with few considerations on the allocation of real-time tasks. To address this issue, in this study, we focus on energy-aware scheduling for real-time tasks in Clouds and propose a novel scheduling algorithm ERES. The ERES integrates the elasticity of Clouds into real-time task scheduling, i.e., ERES strives to reduce the energy consumption of system while guaranteeing the QoS of real-time tasks. If some real-time tasks cannot be finished before their deadlines on current active resources, it dynamically adds new virtual machines (VMs) to enhance the Cloud processing capability although more energy will be consumed. Conversely, when monitoring that some VMs are idle for a period, these VMs will be deleted and other VMs will be consolidated dynamically for energy saving. Extensive simulation experiments are conducted to validate the effectiveness of our ERES. The experimental results show that ERES is superior to others and is suitable for energy-aware real-time task scheduling in Clouds.