Today, the amount of raw data available is abundant. As only a small part of this data is in a form fit for further processing, there is many data left to analyze and process. At the same time, cloud services are ubiquitous and allow even small businesses to perform large tasks of distributed data processing without the significant costs required for a suitable computational infrastructure. However, as more and more users transfer their data into the cloud for processing and storage, concerns about data security arise. An extensive review of data security research in today's cloud solutions confirms these concerns to be justified. The existing strategies for securing one's data are not adequate for many use cases. Therefore, this work proposes a holistic security concept for distributed data processing in the cloud. For the purpose of providing security in heterogeneous cloud environments, it statically analyzes a data flow prior to execution and determines the optimal security measurements. Without imposing strict requirements on the cloud services involved, it can be deployed in a broad range of scenarios. The concept's generic design can be adopted by existing data rocessing tools. An exemplary implementation is provided for the mashup tool FlexMash. Requirements, such as data confidentiality, integrity, access control, and scalability were evaluated to be met.