Greening cloud data centers in an economical way by energy trading with power grid

摘要

In this paper, we mainly study the issues of green scheduling for cloud data centers in an economical way by energy trading with the power grid. The data centers can be powered by the self-generated wind or solar energy, or by the energy purchased from the renewable power plant. The renewable energy can be used to power the data centers directly, or stored into ESDs (Energy Storage Devices) for later use, or sell back to the power grid to finance part of the high energy expenditure of the data centers. It is hard to make decisions on the usage of each type of energy considering dynamic resource demand of different types of requests, time-varying and location-varying electricity prices, and intermittent supply of renewable energy in each time slot. We focus on two optimization problems: (1) Minimizing total energy cost through scheduling of requests, servers, and the usage of different energy sources. (2) Minimizing total carbon emissions within the budget of energy cost. We formulate each problem as an optimization problem during the whole period of time. Our simulation is based on traces from real world. Experiment results show that our scheduling methods can significantly reduce the carbon emissions for cloud data centers.