Prediction-Based Optimization Strategy for VM Migration in Cloud Data Centers

Abstract

Reasonable VM migration can effectively improve the energy efficiency within data centers. To
slow down the high energy-consumption within a data center, a novel VM (Virtual Machine)
migration algorithm based on the prediction has been proposed. Its main idea is to leverage “the
second exponential smoothing method” to predict the resource utilization of the hosts in a data
center, and the VMs in the “overloaded” hosts and the “under-loaded” hosts based on this
prediction will be migrated to a suitable destination host. Then, the idle hosts will be switched to
the energy-aware mode or will be closed to reduce the energy consumption. The experimental
result indicates that, in comparison with other energy-aware algorithms, the proposed algorithm
has certain advantages in terms of energy efficiency.