The basic concept of the Internet of Things (IoT), to uniquely identify objects and to create a virtual representation based on technologies of the Internet, can be extended to so called digital object memories (DOMe), by attaching a virtual storage space to each physical object. This allows for collecting all object-related information generated along the life-cycle chain of this object. The research question, how an infrastructure for digital object memories has to be designed is addressed in this article. Primary goal is to identify and develop components and processes of an architecture concept particularly suited to represent, manage, and use digital object memories. In order to leverage acceptance and deployment of this novel technology, the envisioned infrastructure has to include tools for integration of new systems, and for migration with existing systems. Special requirements to object memories result from the heterogeneity of data in so-called open-loop scenarios. On the one hand, they have to be flexible enough to handle different data types. On the other hand, a simple and structured data access is required. Depending on the application scenario, the latter one needs to be complemented with concepts for a rights- and role-based access and version control. We present a framework based on a structuring data model and a set of tools to create new and to migrate existing applications to digital object memories.