NAM: a nearest acquaintance modeling approach for VM allocation using R-Tree

Abstract

ABSTRACT Cloud datacenters on integration with virtualization provide the dynamic and flexible resource provisioning for the computation or processing of the data-intensive applications. To carry out the operations efficiently in virtualized environment, energy consumption has become one of the major challenges. Therefore, optimal virtual machine (VM) mapping to the Physical Machines is required, otherwise power consumption can drastically hike the overall cost. This paper proposes a novel multi-objective approach for allocating the VMs using dynamic data structure R-Tree which is analogous to bin packing problem. R-Tree optimally handles the accommodation of a large number of multidimensional objects without impacting the depth of the tree. The proposed approach tries to pack as many VMs to the host, without breaching their capacities as to increase the profit. The term profit is a multi-valued attribute which includes the count of hosts, service-level agreement (SLA) Violations, Energy Consumption, and cost of the datacenter. CloudSim toolkit is used to conduct the simulation and the results are analyzed, which shows the reduction in energy consumption and SLA Violations. Hence, it provides enough scope for future research.