Efficient execution of computation modules in a model with massive data

Abstract

Models and simulations for analyzing difficult real-world problems often deal with massive amounts of data. The data problem is compounded when the analysis must be repeatedly run to perform what-if analyses. One such model is the Integrated Consumable Item Support (ICIS) Model developed for the U.S. Defense Logistics Agency (DLA). It models DLA's ability to satisfy future wartime requirements for parts, fuel, and food. ICIS uses a number of computation modules to project demands, model sourcing, and identify potential problem items for various commodities and military services. These modules are written in a variety of computer languages and must work together to generate an ICIS analysis.