Single Instruction, Multiple-Data Database System

Current data mining platforms generally consist of a collection of conventional processors that communicate with each other via a high-speed network (a cluster) or via shared memory (a symmetric multiprocessing system). As one adds more processors to either of these systems, the processing time required to complete a particular task (such as a sequence of database queries) does not decrease linearly, because the processors execute asynchronously. This project will develop a synchronous parallel server system for which the total execution time is a monotonically decreasing function of the number of processors, resulting in a querying speed that is several hundred times faster. In Phase I, previously-developed, performance predictor software will be extended to test the execution time of the parallel computing system. Simulation results will be compared with published performance data for current servers. Commercial Applications And Other Benefits as described by the Applicant: The technology should be especially well suited to applications that require fast queries from a large database. Potential applications include: processing collected data by large government agencies (e.g., the Department of Homeland Security, with respect to their efforts to identify threats to the U.S.), tracking inventory in the business community via radio frequency identification (RFID) tags, and DNA sequence comparison for law enforcement.