Authors: Guofu Feng; Rajkumar Buyya
Addresses: Department of Computer Science and Technology, Nanjing Audit University, 77 Beiwei Road, Nanjing 210029, China ' Cloud Computing and Distributed Systems (CLOUDS) Laboratory, Department of Computing and Information Systems, The University of Melbourne, VIC 3010, Australia
Abstract: Cloud computing is distinguished from such conventional computing paradigms as grid computing and cluster computing in that it provides a practical business model for customers to use the resources remotely. It is natural for service providers to allocate the pooled cloud resources dynamically among the differentiated customers to maximise their revenue. This paper addresses the problem of the revenue maximisation through the SLA-aware resource allocation. Firstly, two TSF (Time Service Factor) based pricing models are proposed since TSF is a widely used metric to determine the billings of internet services with variable performance. Then the resource allocation problem is formulised with queuing theory and its optimal solutions are proposed. The optimal solution considers various Quality of Service (QoS) parameters such as pricing, arrival rates, service rates and available resources. Finally, the experiment results, both with the synthetic dataset and traced dataset, are presented. They have validated our optimal resource allocation solutions and shown that our algorithms outperform the related work.
Keywords: cloud computing; service level agreements; SLA; resource allocation; time service factor; TSF; pricing models; pooled resources; revenue maximisation; queuing theory; quality of service; QoS; arrival rates; service rates; available resources.
International Journal of Grid and Utility Computing, 2016 Vol.7 No.1, pp.12 - 21
Received: 19 Oct 2013
Accepted: 02 Jun 2014
Published online: 18 Dec 2015 *