Sustaining Cyberinfrastructure Interoperabililty

Abstract:

The National Science Foundation has initiated multiple cyberinfrastructure projects through the DataNet Parters, Data Infrastructure Building Blocks, Big Data Hubs, EarthCube, and XSEDE programs. Each project builds a component of cyberinfrastructure, such as collaboration environments, repositories, archives, union catalogs, or data manipulation services. The DataNet Federation Consortium (DFC) (datafed.org) has explored the types of interoperability mechanisms that are needed to build research collaboration environments that span the multiple NSF projects. Based on collaborations with each group, three basic mechanisms have proven to be sufficient: 1) Strong Federation: Tightly coupled federations in which shared name spaces are used for users and files along with communications based on well-defined protocols and API interfaces. Trust relationships based on policies play a major role in this federation. 2) Soft Federation: Loosly coupled federation where one system invokes remote services offered by another using service level communications. No name spaces are shared in such a federation. This type of federation is useful when well-defined services are available remotely on the internet. 3) Asynchronous Federation: Weakly coupled federation in which no direct interaction occurs between the systems. This level of federation is very extensible and flexible.

Long-term sustainability can be achieved when the interoperability mechanisms enable interaction between old and new technologies – with the interactions being flexible or rigid depending upon the strength of trust needed between the systems. Sustaining cyberinfrastructure interoperability becomes feasible when the mechanisms enable the capture of the knowledge needed for interaction, without requiring changes to either sets of infrastructure. The DFC project has implemented all three types of federation to link services and systems based on the needs of the users.