CC*IIE Integration: RADII: Resource Aware DatacentrIc CollaboratIon Infrastructure

University Of North Carolina At Chapel Hill, Chapel Hill NC

Investigators

Abstract

Science collaborations today are multi-institutional and multi-disciplinary, requiring processing of large datasets that are distributed among their participants. They also require advanced computing and networking infrastructure to process the data and arrive at results. A fundamental problem with the way these collaborations operate today lies in the fact that they are structured around the ownership of data or necessary pieces of infrastructure, rather than around the talents and capabilities of participating organizations. This project is developing mechanisms for mapping collaborations onto the new type of dynamically configurable, deeply networked institutional cloud infrastructure, where data is a first-class resource. By integrating data and resource management into a single system RADII aims at improving the productivity of research scientists and reducing time to discovery and operating costs. RADII leverages two previous NSF investments: dynamic infrastructure technologies developed for NSF GENI initiative and the iRODS data grid system for maintaining large distributed datasets. The project is developing software tools to represent science collaborations using formal modeling mechanisms, map data processes, computations and storage onto physical infrastructure and flexibly manage the underlying infrastructure to optimize its utilization across multiple collaborations. The project uses several examples of collaborations from genomics to demonstrate how such collaborations can map onto infrastructure in a more flexible and cost effective way compared to today.