CC*IIE: Integration: COSciN: Cornell Open Science Network

Cornell University, Ithaca NY

Investigators

Abstract

In a growing number of fields, scientists are leveraging computational techniques to make discoveries by collecting, integrating, and analyzing large data sets. But while the proliferation of data has created new opportunities for research, the cyberinfrastructure on university campuses has not kept pace. The labs where data are generated are typically not co-located with the compute clusters used to analyze them, but few campus networks optimize for bulk data transfer. Most campus networks offer "best effort" delivery and do not employ advanced traffic engineering schemes that could provide better service for science flows. Part of the reason for this failure to employ advanced traffic engineering is it makes managing the network much more difficult. In order to reduce the management difficulty and realize the full potential of cyberinfrastructure better tools are needed. This project addresses the tool issues by transitioning prototype implementations of Frenetic which provides a high-level language for expressing and composing network policies, and SoNIC which is a toolkit building precisely controlled network interfaces and associated measurement tools. This work will be done in the operational context of a new research network between Cornell University's main campus in Ithaca and its medical campus in New York City. As a driving application, the project is focusing on the efficient transfer of genomic sequencing datasets, which often run to tens of terabytes and are rapidly growing in size. The project gives scientists a practical tool for transferring data over the wide-area network while transitioning the Frenetic and SoNIC systems from prototype implementations into high-quality open-source systems.