SPRUCE system permits emergency supercomputing

Published 22 November 2006

Scientists need access to TeraGrid computers in order to predict dangerous weather patterns; once notified by an emergency call, computers and servers linked into the SPRUCE system preempt less-critical work

Establishing post-disaster computing services is one of the critical steps in any business continuity plan. This is as true for financial services companies as it is for a small retailing concern, and our readers know that there exists no end of remote access, data storage, and mirroring companies interested in lending a hand. These companies, though they perform a valuable service, are however limited to securing services for standard business needs, however vast those may be in this day and age. Those that rely on supercomputers have so far been out of luck, until now. A new software system under development by the Argonne National Laboratory and University of Chicago bridges the gap by supporting urgent computing on both traditional supercomputers and distributed computational grids.

Intended to provide computational resources quickly for emergency applications affecting public health, safety, and security, the Special Priority and Urgent Computing Environment (SPRUCE), system provides right-of-way tokens that can be activated either automatically via Web services or manually from a Web-based portal. Once notified by an emergency call, computers and servers linked into the SPRUCE system can respond by preempting other jobs or running them immediately after the emergency job completes. This could be critical in permitting time-sensitive analyses of a fast-evolving weather emergency that requires immediate access to such facilities as the TeraGrid supercomputers at the University of Chicago. “Severe weather prediction can be computationally intensive and naturally the workload is unpredictable; we need access to supercomputers quickly when severe storm forecasts are required,” said professor Kelvin Droegemeier. “With SPRUCE, we can rapidly get resources at some of the nation’s largest supercomputer centers.”

The SPRUCE system is currently deployed on NSF TeraGrid resources at five sites: The University of Chicago/Argonne National Laboratory, National Center for Supercomputing Applications, Purdue University, San Diego Supercomputer Center and Texas Advanced Computing Center.

-read more in this ContinuityCentral report