期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2014
卷号：68
期号：3
出版社：Journal of Theoretical and Applied
摘要：Cloud computing is an on-demand resource provisioning technology and server virtualization act as a driving force of cloud. Virtualization consolidates multiple physical machines into one machine, thereby cut cost and improves efficiency of data center. However, as all virtual machines (VM) share the same physical resources, contention for shared resources cause significant variance in observed system response time and throughput. Diverse and unpredictable workloads in cloud environments, needs resource allocations to be continuously optimized to ensure the hosted services meet their service level objectives (SLO). However, the current VMM algorithms are more oriented with providing fair access to the VMs; Lack the ability to adaptively determine the effects of changing resource allocations on the performance of the hosted IT services. Furthermore, as hardware getting evolved and multi core processor technology has increased density of processor cores in a computer at a faster rate, effective usage of the resources becomes a great challenge to software. This is a major bottleneck in cloud applications where performance plays a vital role for user acceptance. Taking this all into account, the paper propose a novel system using meta-heuristic combinatorial search techniques that automatically regulates the VMM CPU scheduler related to the applications on-the-fly with dynamic changes in the environment to maximize throughput and minimize response time. We used this resource allocation algorithm in an evaluation, consist of various scenarios with synthetic workloads. Simulation based results indicate that proposed model improves CPU utilization and make the best tradeoff between resource utilization and performance by 2% on average and up to 6% compared to the default VMM scheduler configurations. The proposed model discussed in this paper can readily be extended to a multi-tier cloud computing environment applications to reduce the overall performance delay.
