University of Maine.

Challenge The University of Maine System has supported a number of research programs across multiple disciplines with a HighPerformance Computing (HPC) landscape since 2012. Each new year brings larger research projects with corresponding increases in data storage demands. For example, one ongoing project, which generates high-resolution ocean models to map climate change, requires half a petabyte of data by itself. Another, employing deep learning to help detect tumors, requires a single directory with over two million files. In some cases, individual files from research teams weigh in at 20TB. The university’s storage architecture was beginning to buckle under the strain of these diverse and demanding data workloads. The existing storage systems were difficult to scale up, and with tight budgets the university could not afford to rip-and-replace the entire infrastructure.