Pittsburgh Supercomputing Center 

Advancing the state-of-the-art in high-performance computing,
communications and data analytics.

networking

21st Century Data Highway

PSC Advanced Networking Group Expands Networking Capabilities for Region, World

In the era of “Big Data,” the challenge of moving the vastly expanded data volumes created and needed by today’s researchers has become central.

The old network—the equivalent of an overcrowded two-lane road—is giving way to a more flexible, software-defined network that manages itself and talks with users to help them work smarter. PSC has contributed to expanding and managing the hardware “lanes” and “merges” so that large-scale users can avoid the traffic jams—and just as importantly, avoid creating them.

The Web10G collaboration between PSC and the National Center for Supercomputing Applications won an NSF grant to create a “dashboard” for networking users. Employing network data extracted by earlier Web10G work, the tool will allow nontechnical users to spot network slowdowns and report them to system administrators for repair.

Thanks to another NSF grant, PSC has “widened the highway,” activating the Pittsburgh region’s first 100-gigabit-per-second Internet2 connection. The improved Three Rivers Optical Exchange (3ROX) connection makes it possible to move data 10 times faster than the previous fastest research and industrial connections—and 5,000 times faster than the fastest home Internet connections.

The newest NSF networking grant is for DANCES.A system of upgraded user-end hardware and innovative software, DANCES will allow high-volume users to schedule existing network resources.The idea is to prioritize routing and scheduling for the biggest jobs, creating virtual direct connections that avoid traffic jams.

We’ve found that a lot of network users either have unrealistically high or unrealistically low expectations. With Web10G, we’re going to automate that process to let users know what’s reasonable and then help them get the performance they need. —Chris Rapier, network applications engineer

DANCES will help supercomputing users employ network resources more efficiently and effectively by scheduling and managing the movement of their data. —Kathy Benninger, DANCES principal investigator