Load Balancing for Virtual Resources Management in Data Center

Abstract

Cloud computing is an come out tendency in the field of Information Technology and load balancing repairs the dispersal of workloads across various computing resources. Load balancing purpose to improve resources manipulation, minimize action time, maximize throughput, and avoid overburden of any single resources. By using of virtualization technologies in the datacenter, the cloud allows organizing assets management. Since the hotspots (i.e., overloaded machines) can reduce the performance of these tasks, virtual machine migration has been utilized to perform load balancing in the datacenters to eliminate hotspots and guarantee Service Level Agreements (SLAs). To address this issue, we propose a Tranche Markov Prediction load balancing scheme. The VM migration algorithm in the scheme aims to minimize the migration cost for load balancing considering the network topology and improves throughput, memory utilization, migration time, execution time and, the worst performance the system, performance of physical machines, frequency, decrease performance degradation and energy consumption could experience from the hotspots. We investigated the distribution of VM demands in a large-scale, and evaluate the impact of different distributions of workloads on the performance of load balancing.