Energy-aware scheduling of virtual machines in heterogeneous cloud computing systems

• Published in: Future generation computer systems 2017-09, Vol.74, p.142-150
• Main Authors: Duan, Hancong, Chen, Chao, Min, Geyong, Wu, Yu
• Format: Article
• Language: English
• Subjects: Cloud computing; Energy efficiency; Fractal prediction model; Resource allocation; Resource-intensive applications
• ISSN: 0167-739X; 1872-7115
• DOI: 10.1016/j.future.2016.02.016

With the rapid development of cloud computing, how to reduce energy consumption as well as maintain high computation capacity has become a timely and important challenge. Existing Virtual Machines (VMs) scheduling schemes have mainly focused on enhancing the cluster resource utilization and reducing power consumption by improving the legacy “bin-packing” algorithm. However, different resource-intensive applications running on VMs in realistic scenarios have significant effects on the system performance and energy consumption. Furthermore, instantaneous peak loads may lead to a scheduling error, which can significantly impede the energy efficiency of scheduling algorithms. In this paper, we propose a new scheduling approach named PreAntPolicy that consists of a prediction model based on fractal mathematics and a scheduler on the basis of an improved ant colony algorithm. The prediction model determines whether to trigger the execution of the scheduler by virtue of load trend prediction, and the scheduler is responsible for resource scheduling while minimizing energy consumption under the premise of guaranteeing the Quality-of-Service (QoS). Through extensive analysis and simulation experiments using real workload traces from the compute clusters of Google, the performance results demonstrate that the proposed approach exhibits excellent energy efficiency and resource utilization. Moreover, this approach offers an effective dynamic capacity provisioning model for resource-intensive applications in a heterogeneous computing environment and can reduce the consumption of system resources and energy when scheduling is triggered by instantaneous peak loads. •An efficient prediction model based on fractal mathematics is developed.•An improved ant colony algorithm for optimizing energy consumption is proposed.•The proposed approach shows excellent energy efficiency and resource utilization.