Migration Cost Aware Mitigating Hot Nodes in the Cloud

Cloud enables dynamic resource sharing among different applications. Live virtual machine (VM) migration is an important way to reconfigure workloads in the Cloud. However, reconfiguration cost using live VM migration is always ignored. In fact, migration cost may vary significantly for different workloads due to diverse characteristics. In this paper, we present a quick and efficient method to make migration decisions to mitigate hot nodes. The process of live VM migration is theoretically analysed. A quantitative model is figured out to estimate migration costs for applications. We implement system prototype based on Xen 3.2.0. The experimental results show that our approach can fast mitigate hot nodes in the Cloud and effectively reduce migration cost.