This work introduces an application of simulation-optimization techniques to the emerging field of green internet computing. The paper discusses the relevance of considering environmental factors in modern computing and then describes how simulation can be combined with scheduling metaheuristics in order to reduce the expected time needed to complete a set of tasks in a server under the realistic assumption of stochastic processing times. This, in turn, allows for a reduction in average energy consumption, which makes the computing facility more efficient from an environmental perspective. Some experiments have been carried out in order to illustrate these potential savings.