Virtual Machine Placement using Enhanced Scheduling and Load Rebalancing using Hybrid Algorithms based on Multi-Dimensional Resource Characteristics in Cloud Computing Systems

VM placement is defined as the process of mapping the VM requests to the PMs, according to the availability of resources in these hosts. In this paper we present an efficient Virtual Machine Placement System called VMP-LR consists of three main components. First, the Resource Request Handling Component creates VM queues for each data centre using Queuing Algorithm based on Multi-dimensional Resource Characteristics. Second, the VM placement Component uses traffic and load aware Scheduling Algorithms to map VMs to PMs efficiently. These algorithms are also fine tuned to efficiently handle high, medium and low resource requests. Third, Load Monitoring Component performs monitoring the VMs periodically in terms of the selected resources and when usage of the PM drops below a threshold, performs rebalancing using an algorithm. This algorithm is based on ant colony optimization and can avoid over and under utilization of resources. All these components are interconnected and will be monitoring continuously by the VM manager. The proposed VMP-LR was deployed on an open source simulation framework called CloudSim. In order to analyze the efficiency of the proposed algorithms, several experiments were conducted and the results were compared with existing conventional solutions. The proposed algorithm utilizes minimum number of physical servers for hosting the set of VMs, which also reduces the energy consumption of the datacenter and it achieved high resource utilization rate by the way of using minimal number of physical servers.