INTELLIGENT CLOUD VM PLACEMENT FOR MINIMIZING IMAGE RETRIEVAL & COMMUNICATION OVERHEAD

Abstract

This study represents a heuristic technique focused on optimizing the deployment of a virtual machine (VM) in cloud data centers focusing on minimizing image search and communication expenses. It presents three new algorithms: PM clustering, VM division and VMPM mapping. The aim of the PM clustering is to reduce the overall communication operation of the longest communication distance between physical machines (PMS) and their aggregation to improve communication efficiency. The VM distribution uses a gross assignment strategy for clusters of analogous VM, increases image search performance and minimizes communication between VM. VM-PM mapping ensures optimal alignment of virtual machines with physical hosts according to the requirements for network and resources topology. Kmeans clustering is used to increase the location by taking into account the distance, search costs and communication. The AWS Cloud implementation shows the usability of real world contexts using EC2 instances and VPC architecture for testing and verification. This complete technique improves VM location, optimizes resource use and significantly reduces operating costs in cloud infrastructures