Auto-ID technologies allow capturing the time and location of products in the supply chain for tracking and tracing. This paves the way for a variety of business applications, such as anti-counterfeiting, pedigree, and genealogy, which analyze the trace history of products to detect patterns or anomalies in the supply chain. While these applications have gained considerable interest recently, further work is needed towards integration of event data from heterogeneous auto-ID nodes in order to obtain the complete trace history for products of interest. As a first step, we perform an architectural study on interoperable auto-ID systems and present the results in this paper. We first review established techniques for data integration and data sharing as well as relevant industrial efforts. We then clarify the requirements that need to be addressed by an auto-ID network. Finally, we discuss four possible architecture alternatives for implementing interoperability in such a network and comparatively evaluate the approaches according to the identified requirements