Abstract:

ABSTRACT This work addresses the challenge of optimizing both electricity consumption and maintenance costs in Cloud Data Centers (CDCs) by intelligently managing the power states of servers. Specifically, we focus on the allocation of virtual machines (VMs) based on their CPU and memory requirements over multiple time slots. The model incorporates electricity usage based on VM processing loads, data transfer between VMs, and VM migration across servers. Furthermore, a material-based fatigue model is used to estimate CPU maintenance costs arising from temperature fluctuations due to dynamic power variations. To solve this joint optimization problem, we propose a novel algorithm named Maintenance and Electricity Costs Data Center (MECDC). Experimental evaluations on real CDC scenarios demonstrate that MECDC significantly outperforms baseline algorithms that focus solely on load balancing or energy efficiency, offering a more holistic and cost-effective solution for data center operations.

Description:

.