Scientific research is based upon the ability to compare theory and simulations with experimental and observational data. The data are assembled in reference collections to enable comparison of future analyses with the current state-of-the-art understanding. The reference collections are published for use by the entire scientific discipline. This process is used by all science disciplines to document scientific progress and facilitate exchange of knowledge. We present an approach that has been used to implement data grid technology that is used internationally in support of data sharing, data publication, data preservation, and data analysis.