Resource sharing between multiple tenants is a key rationale behind the cost effectiveness in the cloud. While this resource sharing greatly helps service providers improve resource utilization and increase profit, it impacts on the service quality (e.g., the performance of consumer applications). In this paper, we address the reconciliation of these conflicting objectives by scheduling service requests with the dynamic creation of service instances. Specifically, our scheduling algorithms attempt to maximize profit within the satisfactory level of service quality specified by the service consumer. Our contributions include (1) the development of a pricing model using processor-sharing for clouds (i.e., queuing delay is embedded in processing time), (2) the application of this pricing model to composite services with dependency consideration, (3) the development of two sets of service request scheduling algorithms, and (4) the development of a prioritization policy for data service aiming to maximize the profit of data service.