Dynamic Workload Migration Over Backbone Network to Minimize Data Center Electricity Cost

Abstract

As more organizations adopt cloud services, energy consumption in data centers (DCs) keeps increasing. Today, information and communication technology (ICT) has become a major consumer of energy worldwide. A large portion of ICT energy consumption is used to power servers running in DCs and the network they use to communicate. In this paper, we consider that energy cost at a particular DC is often related to the electricity price regulated by independent system operators/regional transmission organizations. As these prices vary in time and depend on the geographical locations of the DCs, recent studies have shown that the spatio-temporal variations of electricity price can be exploited to reduce electricity cost. In particular, as workloads tend to change, often unpredictably, adaptive workload placement and migration can help to serve the workloads in regions with lower electricity costs. While most prior works consider a quasi-static scenario with known workload patterns, this paper proposes dynamic workload-aware algorithms which exploit the spatio-temporal variations of electricity costs to minimize the energy cost in ICT. Although prior studies focused on power consumption from power consumers such as servers, cooling systems, etc., recent studies have shown that the network elements consume a significant portion of the energy. Hence, while reducing DC energy cost, this paper also considers electricity cost of the backbone network. Algorithms introduced in this paper use dynamic request re-routing and live virtual machine (VM) migration to move workloads to DCs with lower electricity cost. We consider VM migration cost (including electricity cost at backbone network nodes), bandwidth constraints for migration, VM consolidation, constraints from service level agreement, and administrative overhead of VM migration. Our simulation studies show that the proposed algorithms reduce operational cost of DCs significantly.