INRA and IRSTEA, 2 French research organizations, have joined together to become INRAE, a world class institute for research on agriculture, food, and environment, mainly funded by public resources. INRAE has recently set out its principles of "data governance", to cover all the processes required to manage and enhance data sharing on the basis of ethics, legal, economic, technical, and scientific policy criteria. Roles and responsibilities of the actors are defined to ensure a smooth and sustainable decisional process. A "data governance" charter has been written to indicate who decides, which data, and how they are opened according to good practices such as the FAIR principles. With this document all scientists, administrative support, and data managers involved in the data life cycle have a shared understanding of the rationale guiding the global data framework. We address here the question of "what can be a data governance framework at an institutional level to support both data management, sharing, and reuse. We've defined 4 key principles at the foundation of data governance: (1) data must be shared and reused while observing the values of science, (2) data must be managed in order to make it F.A.I.R., (3) data should be "as open as possible, as closed as necessary", and (4) open data contributes to innovation and value creation for society. The key rationale of these four principles is that they build a consistent "system" for guiding the decision, as it requires a careful evaluation of all of them. These principles were complemented with a decision-making chain including the main actors. This "data governance schema" has been built thanks to a participative approach. It is the result of several rounds of discussion and consultation with different groups of people having different views on data and different levels of responsibilities, including interviews of scientific project leaders on a dozen significant case studies. It took 1 year to establish our first guidelines. This schema is today in the process of implementation. Improvements will certainly come after the first real-life usage. However, we believe that these four principles of governance are generic enough to be applied to many research organisms, only the internal organization or process should differ according to the culture and the regulation context.