Software as a Service is a software deployment method. Vendors employing it license customers to access the software on demand over the Internet. It allows that computing resources are shared between customers without them noticing. It also respects that costumers have different requirements by enabling them to adjust the properties of the application. This customization ranges from altering application internal processes to specifying individual service level agreements. Software as a Service vendors are therefore able to extend their target market drastically since the demands of a much larger consumer group can be met. To fulfill the additional characteristics of Software as a Service a number of challenges have to be addressed. Customizable qualities such as availability or security are strongly interleaved with the computing environment. Since fine nuances of these qualities are required for good consumer targeting multiple computing environments have to be utilized. Integration and management of these heterogeneous environments is vital for the success of a Software as a Service vendor. Within the scope of this thesis integration of these environments and the load balancing of users according to their requested service level agreements is enabled. An abstract data model is defined describing the resources constituting the managed system as well as its users. Optimization of the user distribution is performed based on the information contained in this model. A system architecture able to reflect the optimization results is also introduced. It allows automatic provisioning of needed resources in the managed computing environments and reflects the distribution of users by routing their requests to assigned resources.