Utilization based Artificial Bee Colony (UbABC) Optimization Algorithm for Virtual Machine Placement in Cloud Data Centers

Abstract

Cloud data centers provide services on demand to the customers through virtualization, a key technology that enables the services to be provided through Virtual Machines. Cloud data centers that host a large number of services leverage virtualization techniques to improve resource utilization and to reduce cost. The key factor to the success of the virtualization technique is to have a Virtual machine placement algorithm that optimally places the Virtual Machines on the Physical Machines of the Cloud Data center. This paper proposes a Utilization based Artificial Bee Colony Optimization algorithm that explores the placement solution based on the resource availability and utilization. The objectives of the proposed Virtual Machine placement threefold: 1. To enable the cloud service providers optimally manage a large number of heterogeneous workloads from multiple users; 2. To handle different processing requirements under different computational loads and infrastructure scales; 3.  To improve the resource utilization of the active hosts.  The VM allocation using the UbABC algorithm maps the VMs dynamically to the PMs and also takes care of the migrations when the underlying PM is overutilized. The optimized UbABC migrates the VMs from the overutilized to the underutilized PMs based on a certain threshold, which is determined by the Decision Tree Regression algorithm