Cost-aware workload consolidation in green cloud datacenter

Server consolidation is important for better resource utilization and efficient energy saving for cloud datacenters which host thousands of virtual machines (VMs) to support multitenant applications. The typical approach is to migrate VMs and reallocate workload among different servers in a way to minimize the total number of servers used. However, most existing works for server consolidation focus mainly on how to reduce the number of active servers and do not account for the migration overhead incurred to the applications on the migrating VMs (such as downtime). In this paper, we propose an adaptive mechanism to schedule VM allocation in cloud datacenters. Our solution takes into account the resource utilization and migration overheads, and adaptively allocates each VM to servers based on the estimated saturation level. As a result, the quality and overhead of consolidation is balanced and the total cost is minimized. The simulation results show that our mechanism could increase the average utilization on servers by up to 97% while reducing the total migration cost by about 60%, as compared with existing solutions.