MRI: Acquisition of a Shared Scalable Research Storage System

George Mason University, Fairfax VA

Investigators

Abstract

This project, acquiring a high-end scalable network storage system, aims to integrate an innovative storage architecture (MEMORI) into the existing research computing infrastructure at the institution to address the needs of both traditional and non-traditional HPC researchers. The proponents will then instantiate that architecture with the acquisition of the high-end scalable network storage system. The instrumentation will be managed centrally, but shared with the entire research community. The instrumentation enables a variety of projects in cyber-security, mental health, climate dynamics and social systems, all of which positively benefit the health and economic well-being of the country. In particular, the storage system is designed to support shared access by sister institutions, since GMU is a member of the 4-VA university consortium that provides state-level infrastructure promoting resource sharing. Data access by potential outside stakeholders will also be enhanced. The current research computing infrastructure is unable to support the growing storage needs for the wide variety of current research projects involving elements of "big data." As a result, individual projects are limited in their ability to store, mine, visualize and share large data sets. Rather than forcing research projects to deal with these limitations on an individual and ad hoc basis, the researchers will take advantage of recent developments in storage technologies to provide a shared storage server capable of supporting current storage needs and scalable in a cost-effective manner for future needs. The proposed architecture is specifically designed to support research projects in the humanities and social sciences, as well as the more traditional science and engineering disciplines. Furthermore, centrally managing the instrumentation creates a secure environment for multi-university shared access, appropriate control of sensitive data, and a reliable framework for data management plans, backups, and archiving. In addition, a variety of departments offer courses in data mining, data analytics, and data visualization. These courses are currently limited in their ability to support datasets of significant size, and will greatly benefit from the proposed system. The library services will be significantly enhanced by the instrumentation. Fast network access to the system provides the ability to store local copies of large databases for text mining research by faculty and students, as well as visualization capabilities.