CINET - A Cyber-Infrastructure for Network Science
Abstract
CINet is a cyberinfrastructure middleware to support Network Science. The National Research Council defines Network Science as “the study of network representations of physical, biological, and social phenomena leading to predictive models of these phenomena.” This middleware will give Network Scientists access to an unparalleled computational and analytic environment for research, education and training. There is a growing importance of networks in diverse fields. Recent results in the emerging field
Network Science enhance our understanding of physical and social systems. Advances in computing and information systems provide motivation for developing this middleware. Network Science often deals with very large graphs and time-consuming computations on those graphs that require more computing power than is available on the typical desktop. Scientists working with networks are often not expert computer users; this has hampered the adoption of High Performance Computing (HPC)-based modeling methodologies and environments. By harnessing new cloud-based resources, such as Magellan and FutureGrid, in an easily accessible manner, the proposed work will enable Network Science researchers to tackle larger, more complex problems.
The project vision is to provide researchers, analysts and educators interested in Network Science with an easy-to-use cyber-environment that is accessible from their desktop and integrates into their daily work. A key goal is to greatly expand the size of networks that are routinely studied from hundreds or thousands of nodes to hundreds of millions of nodes. It will leverage the technology, data and experience of a multi-institutional team (Virginia Tech, Indiana University, University of Houston-Downtown, University of Chicago and Argonne National Laboratory) in this area. The idea for the project grew out of frequent requests for access to large synthetic populations and associated models generated at Virginia Tech, as well as the difficulty VT researchers faced acquiring access to needed datasets.
Please see http://ndssl.vbi.vt.edu/cinet for more information.
Intellectual Merit
CINet will enable fundamental changes in the way researchers study and teach complex networks. The use of state-of-the-art computing resources to synthesize, analyze, store and reason about large networks will enable researchers and educators to study networks in novel ways. It will enable educators to harness HPC technologies to teach Network Science to students spanning various academic levels, disciplines and institutions. It will be designed for scalability, usability, extensibility and continuity. The investigators also will advance the fields of digital libraries and grid computing by stretching them to address challenges related to Network Science.
Broader Impact
The investigators will launch a comprehensive education and outreach plan comprised of short courses, workshops at important conferences and focused user group meetings to provide a path towards adoption of the cyberinfrastructure and suggest user-guided improvements. The educational plan includes high school students to Ph.D. candidates, students from minority and under-represented groups, and students at smaller institutions that often do not have easy access to HPC resources. Network Science
is a transdisciplinary topic; the proposed platform will foster multi-disciplinary and multi-university research and teaching collaborations, including areas which have not traditionally made use of HPC resources such as the social sciences.
Use of FutureGrid
FutureGrid will be one of the compute resources available for use by CINet. It will be used by CINet to compute network measures, such as clustering coefficients or degree distribution, on small to large graphs by several graph algorithm systems. The VMs will
contain the graph measure execution environment, and will be provided details of the specific measures and graphs to run by the CINet system. CINet currently supports traditional HPC clusters, and support is being added for grid based systems such as FutureGrid. CINet is an open access system, meaning that anyone will be able to use the system. Select parts of the system will be made available as open source, including the resource brokers.
Scale Of Use
FutureGrid will be one of the compute resources available for use by CINet. One small VM will need to be online continuously to run the local resource broker. This broker will start and stop VMs in response to the compute requests that are sent to FutureGrid. The
central resource broker, run on Virginia Tech resources, can be tuned to send an appropriate number of compute requests to FG so as to achieve the desired utilization.
Futuregrid is a resource provider for