Workshop on Federating Computational Resources

Princeton University, Princeton NJ

Investigators

Abstract

There is increasing demand among researchers and production system architects to combine (federate) compute, storage, and network resources from multiple sources (e.g., the organization?s own resources, their partners? resources, commercial and academic clouds, programmable network substrates). Various proprietary and experimental systems have taken the first steps to demonstrate the potential effectiveness of such federations, but substantial concerns remain about security, interoperability and management. In many ways, the situation resembles what emerging networks faced at the dawn of the Internet. Consolidation seems certain, but there is a lack the right architectural framework, where new models must contend with a quickly growing base of incompatible production systems. Against this backdrop, this workshop focuses on issues related to federating resources from multiple autonomous organizations into a seamless/ubiquitous resource pool, thereby giving users standard interfaces for accessing the widely distributed and diverse collection of resources they need. This workshop brings together leading network researchers and network research infrastructure developers to discuss specifics of federate (combine) compute, storage, and network resources from multiple sources. The concept for the workshop follows from multiple discussions in the GENI and FIRE initiatives to build federated network test-beds. The workshop expects to develop a common understanding of what it means for autonomous organizations to federate their resources into a seamless/ubiquitous resource pool.