SCIENCE

Indiana University's Stephen Simms has been elected to the Community Representative board seat for OpenSFS. Simms is the Data Capacitor project lead and manager of the High Performance File Systems group. He spearheads IU's development of the Lustre file system, and is the university's representative to OpenSFS.

OpenSFS is a technical organization that promotes collaboration among institutions using open source scalable file systems—such as Lustre—in high performance computing (HPC) environments. IU joined OpenSFS in 2010.

"Our original goal in joining OpenSFS was to help ensure the support and further development of the Lustre file system," said Simms. "Lustre supports seven of the top 10 most powerful supercomputers in the world, and has been an important part of IU's cyberinfrastructure since 2006. We are committed to its future growth, and I'm very pleased and proud to represent the interests of community-level participants such as IU."

Norman Morse, president and CEO of OpenSFS, said, "I'm excited to have Stephen's expertise, experience, and leadership on the board. He brings much to the table when it comes to the development and support of this highly valued tool."

In 2005, IU received a National Science Foundation grant for the Data Capacitor, a high-speed, high-bandwidth research storage system that uses Lustre. Lustre has since become the site-wide file system for IU's HPC environment, making it critical to IU scientists and nationwide researchers who use IU HPC resources.

IU has pioneered the development of Lustre across the wide area network. In 2007, IU developed the open source user ID mapping code that currently enables Lustre WAN file systems on the TeraGrid, the Data Capacitor WAN file system, and the Albedo wide area file system.

In 2007, a team led by Simms and IU used the Data Capacitor and the Lustre file system to win first place in the Supercomputing Bandwidth Competition at the annual Supercomputing Conference. Competitors were challenged to create methods for fully utilizing a high-speed network path to support end-to-end network applications running across a grid that included the conference's exhibit floor and the participant's home institutions using production networks.

In 2008, the IU Data Capacitor team started using Lustre in production across wide area networks as part of a service providing 350TB of storage for collaborative research projects mounting file systems across the TeraGrid network and other national high speed networks. This service has proven valuable to many types of scientific research including life sciences, environmental science, and astronomy—all of which require teams of scientists to move, share, and access very large data sets among distant and remote locations.

In 2010, IU launched a partnership with Whamcloud to improve upon IU's pioneering work in the use of Lustre over wide area networks. With help from Whamcloud's team of expert Lustre networking technicians, the IU team is working to increase the performance of their Lustre service.