Title: Efficient resource provisioning for virtual clusters on the cloud

Authors: Shao-Jui Chen; Chia-Ching Chen; Hsiang-Lun Lu; Wei-Jen Wang

Addresses: Department of Computer Science and Information Engineering, National Central University, Taoyuan City, Taiwan ' Department of Computer Science and Information Engineering, National Central University, Taoyuan City, Taiwan ' Department of Computer Science and Information Engineering, National Central University, Taoyuan City, Taiwan ' Department of Computer Science and Information Engineering, National Central University, Taoyuan City, Taiwan

Abstract: A virtual cluster could consume a considerable amount of network bandwidth due to its internal communication demands. Since communication among VMs on the same physical machine is done through emulation, it does not use any network bandwidth. Therefore, placement of virtual clusters should utilise this fact to reduce unnecessary bandwidth consumption. To achieve this goal, one has to consider the constraints of the physical machines, as well as the communication relationship among the virtual machines of the virtual cluster. In this paper, we propose to use a monitoring/profiling mechanism to track and to classify virtual clusters. The information is then used by a resource provisioning mechanism to place virtual machines on physical machines, such that unnecessary network bandwidth consumption is reduced. Our experimental results show that, the proposed mechanism significantly outperforms the default VM scheduler provided by OpenStack.

Keywords: virtual clusters; virtual machines; VM placement; cloud computing; resource provisioning; bandwidth consumption; cluster tracking; cluster classification.

DOI: 10.1504/IJSTM.2017.081876

International Journal of Services Technology and Management, 2017 Vol.23 No.1/2, pp.52 - 63

Received: 19 Mar 2015
Accepted: 11 Mar 2016

Published online: 30 Jan 2017 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article