A Scheduling Strategy for Global Scientific Grids - Minimizing Simultaneously Time and Energy Consumption

Grid computing has consolidated itself as a solution able of integrating, on a global scale, heterogeneous resources distributed geographically. This fact has contributed significantly to increase the IT infrastructure. However, all this computer power results in a lot of energy consumption, raising concerns not only with respect to economic aspects, but also regarding environmental impacts. Current data shows that the information technology and communication industry has been responsible for 2% of the carbon dioxide global emission, equivalent to the entire aviation industry. This paper proposes a biobjective strategy for resource allocation on global scientific grids, considering both energy consumption and execution times. An algorithm is presented which generates the minimal complete set of Pareto-optimal solutions in polynomial time. Computation experience is reported for three distinct scenarios.