An Energy Aware Unified Ant Colony System for Dynamic Virtual Machine Placement in Cloud Computing

Xiao-Fang Liu,Zhi-Hui Zhan,Jun Zhang

doi:10.3390/en10050609

Xiao-Fang Liu, Zhi-Hui Zhan + Show 1 more

Open Access

https://doi.org/10.3390/en10050609

Copy DOI

Abstract

Energy efficiency is a significant topic in cloud computing. Dynamic consolidation of virtual machines (VMs) with live migration is an important method to reduce energy consumption. However, frequent VM live migration may cause a downtime of service. Therefore, the energy save and VM migration are two conflict objectives. In order to efficiently solve the dynamic VM consolidation, the dynamic VM placement (DVMP) problem is formed as a multiobjective problem in this paper. The goal of DVMP is to find a placement solution that uses the fewest servers to host the VMs, including two typical dynamic conditions of the assignment of new coming VMs and the re-allocation of existing VMs. Therefore, we propose a unified algorithm based on an ant colony system (ACS), termed the unified ACS (UACS), that works on both conditions. The UACS firstly uses sufficient servers to host the VMs and then gradually reduces the number of servers. With each especial number of servers, the UACS tries to find feasible solutions with the fewest VM migrations. Herein, a dynamic pheromone deposition method and a special heuristic information strategy are also designed to reduce the number of VM migrations. Therefore, the feasible solutions under different numbers of servers cover the Pareto front of the multiobjective space. Experiments with large-scale random workloads and real workload traces are conducted to evaluate the performance of the UACS. Compared with traditional heuristic, probabilistic, and other ACS based algorithms, the proposed UACS presents competitive performance in terms of energy consumption, the number of VM migrations, and maintaining quality of services (QoS) requirements.

Highlights

Cloud computing is a large-scale distributed computing paradigm in which customers are able to access elastic resources on demand over the Internet by a pay-as-you-go principle [1,2,3]
Due to the good performance of ant colony system (ACS) on the static VMP problem presented in our previous work [33], we developed an ACS based unified algorithm, unified ACS (UACS), for the dynamic virtual machine placement (DVMP)
best-fit decreasing (BFD) is widely adopted for DVMP with a consolidation ratio of 11/9 and is used as a baseline

Summary

Introduction

Cloud computing is a large-scale distributed computing paradigm in which customers are able to access elastic resources on demand over the Internet by a pay-as-you-go principle [1,2,3]. Cloud computing facilitates services at three different levels, including Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Software as a Service (SaaS), and four deployment models, including private cloud, community cloud, public cloud, and hybrid cloud [4]. Reports show that energy consumption has occupied a significant proportion of the total cost of data centers [7]. Energy efficiency is becoming a challenge in data center management [8,9,10,11]. Virtualization is adopted for abstraction and encapsulation such that the underlying infrastructure can be unified as a pool of resources and multiple applications can be executed within isolated virtual

Methods

Results

Conclusion