Distributed and replicated data management in the Cloud is governed by the CAP theorem. As a consequence of this theorem, many Cloud providers sacrifice data consistency for a high degree of availability. However, with the pay-as-you-go cost model of the Cloud where the use of each resource is charged at a very fine-grained level, the costs of an application are introduced as additional parameter for optimization. This can include the infrastructure costs (i.e., the resources needed for guaranteeing a certain degree of data consistency), but also inconsistency costs (i.e., costs that incur when a relaxed level of consistency is provided). In this work, we re-consider existing protocols for distributed data management and develop new protocols for cost-aware, dynamic and adaptable replicated data management in a Cloud environment.