Route Aware Virtual Machine Migration in Cloud Datacenter

Cloud Datacenter holds huge number of servers hosting infrastructure, compute and storage as service. Servers are connected by switches and high performance communication links. They execute wide variety of applications such as e-commerce, social networking and scientific computing. Such compute intensive applications run on virtual machines hosted on cloud servers. Virtual Machine Migration is one promising technique that aids the cloud datacenter to provide energy efficient, fault tolerant service to the cloud customers. They move virtual machine running in one server to another server for the purpose of consolidation and fault tolerance. However, the resource requirement of the source virtual machines is reserved in the target server using the current static utilization. In dynamic cloud environment, it may lead to sudden rise in resource requirement and cause resource starvation in destination server. To address this issue, this paper proposes resource and route aware live virtual machine migration technique. In this technique, the dynamic resource utilization of the servers are predicted using combined forecasting technique. Based on the prediction, the servers are clustered into eight clusters. Then the virtual machines are migrated from overloaded servers to the nearest underutilized server. Simulation is carried out using Google cloud traces. The performance metrics such as migration time, down time and number of migrations are measured. The performance evaluation exhibits that the proposed technique outperforms the state of art technique available in the literatures.