Successful Internet service offerings can only thrive if customers are satisfied with service performance. While large service providers can usually cope with fluctuations of customer visits retaining acceptable Quality of Service, small and medium-sizes enterprises face a big challenge due to limited resources in the IT infrastructure. Popular services, such as justin.tv and SmugMug, rely on external resources provided by cloud computing providers in order to satisfy their customers demands at all times. The paradigm of cloud computing refers to the delivery model of computing services as a utility in a pay-as-you-go manner. In this paper, we provide and computationally evaluate decision models and policies that can help cloud computing providers increase their revenue under the realistic assumption of scarce resources and under both informational certainty and uncertainty of customers? resource requirement predictions. Our results show that in both cases under certainty and under uncertainty applying the dynamic pricing policy significantly increases revenue while using the client classification policy substantially reduces revenue. We also show that, for all policies, the presence of uncertainty causes losses in revenue; when the client classification policy is applied, losses can even amount to more than 8%.