GreenBDT: Renewable-aware scheduling of bulk data transfers for geo-distributed sustainable datacenters

The fast proliferation of cloud computing promotes the rapid growth of datacenters. More and more cloud service providers use geo-distributed green datacenters to support the expanding scale of cloud applications as well as minimize the carbon footprint. In such a geo-distributed green datacenter system, a basic and urgent demand is inter-datacenter bulk data transfer that is usually used for periodic data backup, software distribution, virtual machines cloning, etc. Though many existing research efforts have been made to build green datacenters or provide optimal scheduling for inter-datacenter bulk data transfers separately, still the goal for optimal scheduling of inter-green-datacenter bulk data transfers is being underachieved. This is an important problem, especially when an increasing number of geo-distributed datacenters are powered by renewable energy for reducing energy cost and protecting environment. In this paper, we study the problem of maximizing renewable energy use and minimizing grid energy cost for bulk data transfers between sustainable and green datacenters. We model this problem and propose a heuristic method to solve it. The proposed method is the first to explicitly address the green energy use maximization and grid energy cost minimization problem of inter-green-datacenter bulk data transfers for green and sustainable datacenters in the multi-electricity market environment. Extensive evaluations with real-life network topology, available wind power, and electricity prices show that our method can maximize renewable energy use and bring more energy cost savings over existing bulk data transfer strategies.