A Performance-oriented Resource Allocation Algorithm with Insertion and Duplication for IaaS Clouds

Abstract

The innate characteristics of a cloud computing environment make itself attractive for running applications with large scale data, as a user can utilize a number of high performance computing virtual machine instances without maintenance cost. The performance of executing an application in a cloud computing environment depends on a resource allocation policy that schedules an input task graph onto virtual machine instances. In this paper, we propose a novel resource allocation policy that considers characteristics of an input task graph and virtual machine instances of a cloud computing environment. And also, we propose a new task insertion method that also considers an execution finish time of a child task of the inserted task. Simulation experiments with task graphs from standard task graph project show that the proposed algorithm outperforms conventional algorithms in terms of normalized total execution time.