Manufacturing companies need to be more efficient, productive and achieve new environmental standards while still providing low-cost products with regard to the competition.
A digital production enables large-scale productivity gains, but also requires large amounts of data. As production environments are inherently heterogeneous it is often not easy to record every possible feature. This thesis provides a concept for an infrastructure that allows to record data from very different sources while maintaining a link to product entities. Additionally to this concept, a deployment method is demonstrated that enables the quick deployment of custom applications to the edge to speed up data gathering efforts. This work is done in cooperation with a leading manufacturer of automotive glazing to evaluate the concept and its realization with real data originating from a real environment.