A green-aware virtual machine migration strategy for sustainable datacenter powered by renewable energy

• We propose a green-aware approach for managing VMs in datacenters.
• Models are established for datacenters, VMs and power consumption.
• An optimization problem is formulated to incorporate revenue and costs.
• Three strategies designed to manage VMs dynamically are implemented.
• The proposed approach is proved to be feasible in utilizing green energy.

With the rapid growth of cloud services, huge energy consumption of the underlying large-scale datacenters becomes a major concern of both the resource providers and the society. Datacenter owners are beginning to use renewable energy as extra supply for the devices. In this paper, we design a green-aware power management strategy for such datacenters powered by sustainable energy sources, considering the power consumption of both IT functional devices and cooling devices. Specifically, we make use of energy-aware methods to formulate an overall optimization problem, and try to solve it by combining heuristic and statistical searching approaches. The ultimate objective is to utilize green energy sufficiently while keeping the demand of applications deployed inside the datacenter at an acceptable level. Performance evaluation and simulation experiments are designed upon a simulated testbed, with realistic workload traces and solar energy generation considered, in order to validate the feasibility of our approach. Results show that it can significantly improve the green energy utilization, and achieve the highest overall revenues for the resource provider.