Healthcare is increasingly becoming data-rich with numerous new types of data being generated in heterogeneous distributed environments, from medical and wearable devices to clinical trials and electronic health records. The growing level of eHealth applications, such as mobile healthcare applications for telehealth and remote monitoring, health and wellbeing games, and medical devices that capture health-related data from patients in their environments, is just a few examples in this ever-moving domain. In the present scenario, interoperability and model aggregation across computing infrastructures and heterogeneous data formats have become an increasingly important focus of health informatics. Centralized warehousing of health data is risky, expensive, and damaging in some cases, while skipping the design effort to achieve an interoperable data architecture can yield adverse results in the long run, including insecure connections and embedded document proofs. Achieving semantics-based interoperability by having a shared conceptual model of the involved domains augments the probability of success by providing a detailed vocabulary and grammar for specialization, together with constraints for control. Data aggregation and model interoperability are the double challenges of Health Data Architectures. Clients want to see the big picture, while analysts need to drill down to particular data points to know problems locally. Security and privacy, big data scale, access performance and costing, scalability, and loose coupling among data from different centers and regions interacting with varied rules are the main characteristics and principles that lead to the four abstractions at the Data Center level, the Federation of Ands, the Organization Data Model, and the User-Centered interactivity, with innovative strategies that deal with the right subset of the above properties according to the application area. The latter contains innovative strategies that deserve further testing according to underlying technical changes, such as those concerning the cloud internals. It is time to perform further investigations in the Health Data Architecture.