Today’s shift in the offering method of an enterprise application from a traditionally onpremise product to a managed service within a cloud environment attempts to reduce reduces costs through resource sharing and utilization optimization. To benefit from this advantage, enterprises require a method for implementing elasticity in their traditional applications and enable cloud-aware characteristics. Furthermore different demands from Service-Level- Agreements (SLA) or economical targets must be considered. With enterprise applications not being developed for a cloud environment, optimization is very complex. This thesis will present a generic method that uses the expert knowledge of the enterprise within a feedback loop to enable a dynamic scaling for managed services. This generic method is then being transformed into a concrete architecture and data model. To meet different types of goals, this work outlines different sources of heuristics, their combination and implementation. Finally a prototype implementation located in an enterprise content management setting shows the validity of the solution’s approach and gives further suggestions about improvements and future research topics. Using this method, enterprises can move their applications to any cloud environment, provide their domain specific expert knowledge and provide an elastic enterprise cloud service to save costs for them and their customers.