Application Classification Based on Preference for Resource Requirements in Virtualization Environment

Abstract

Different applications have different preferences for resource requirements. In virtualization environment, if multiple virtual machines hosted on the same server have the same resource requirement preference, performance can be greatly affected for the resource competition between virtual machines. In this paper, we propose an approach to use a feature weighting naive Bayes classifier with Laplacian correction model to classify the applications according to the characteristics of application accessing to CPU, memory, hard disk, and the L2 cache collected using profiling. Based on the application classification, the virtual machines running applications of different types can be deployed on the same physical host. The experiments show that this method can achieve high classification accuracy. And this methods avoid the performance bottleneck due to resource competition to a certain extent.