The National Animal Identification System is a voluntary disease traceability framework released by the USDA. The primary goal of the program is to record the movements of every animal in the U.S. food supply in a massive national database. In the event of a disease outbreak the database would subsequently be searched, exposed animals and premises identified, and the spread of infection contained.

This research is concerned with accurately modeling and analyzing the data processing requirements for such a program. The model consists of two elements; the first being a parallel, Monte Carlo discrete events simulator capable of rapidly creating massive datasets statistically representative of actual NAIS datasets. The second component consists of a parallel algorithm mapped onto an SMP architecture that can rapidly trace an infected animal and identify the network of exposed animals.

The presentation will begin with a brief overview of the NAIS program and the Monte Carlo discrete events simulator along with the datasets created. The remainder of the talk will consist of an illustration of the effective and efficient mapping of the parallel traceability algorithm onto an SMP architecture where a hybrid OpenMP/MPI approach is employed in order to take maximum advantage of the memory hierarchy. The talk will conclude by presenting parallel performance and scaling results of the hybrid implementation including the performance results for a 24 GByte