Service-oriented computing enables organizations to package their offerings as modular, self-contained services. As these organizational capabilities are more commonly distributed, perhaps redundantly, across cloud environments, there is an opportunity to provision services in locations that maximize energy-efficiency. In this paper, we introduce a process, model, and power estimation techniques to facilitate a priori decision support with respect to provisioning software services in context of the server on which they reside.  Favorable results from our experiments demonstrate the ability to estimate the power consumption of web services while differentiating energy-efficiency of redundant instances.