Service-oriented environments are central backbone of todays enterprise workflows. These workflow includes traditional process types like travel booking or order processing as well as data-intensive integration processes like operational business intelligence and data analytics. For the latter process types, current execution semantics and concepts do not scale very well in terms of performance and resource consumption. In this paper, we present a concept for data streaming in business processes that is inspired by the typical execution semantics in data management environments. Therefore, we present a conceptual process and execution model that leverages the idea of stream-based service invocation for a scalable and efficient process execution. In selected results of the evaluation we show, that it outperforms the execution model of current process engines.