Improved Grouping Genetic Algorithm for Dynamic Server Consolidation in Virtualized Data Centers

The advantages of virtualization have made it a key technology in service management of data centers, but the problem of server consolidation which aims to minimize the number of used physical machines, is still a challenging issue. Server consolidation involves virtual machines (VMs) migration which has a major effect on service performance. Existing consolidation algorithms generate unnecessary migrations while minimizing the number of physical servers hosting these VMs. In this paper, we used our previously proposed improved grouping genetic algorithm with a special technique to avoid unnecessary migration for VMs with steady workload. The testing scenarios show encouraging results regarding overall performance with minimum penalty on the number of used servers.