CHARLES RIVER ANALYTICS, INC. — Department of Defense SBIR Phase I: A14-025

Description

Two factors are driving the need for learning and adaptive Soldiers: the rapid development of new technology and a set of continually evolving adversaries. Simulation-based training (SBT) provides a cost-effective alternative to live exercises, but existing virtual environments such as the Military OpenSimulator Enterprise Strategy (MOSES) rely on traditional relational databases (RDBs), limiting them to relatively small numbers of simultaneous users because the data generated quickly outpaces the RDBs? capacities. This prevents SBT from being used for large-scale training exercises. To increase the scalability of SBT, we propose to design and demonstrate the feasibility of Non-Relational Scaling for Virtual Environments (NORSE). NORSE replaces relational databases with horizontally scaling and distributed non-relational database clusters presented as a set of generic services that are agnostic to any specific virtual environment. NORSE includes an instrumented testing framework and a customizable test-data generator, a test harness, which enables real-time measurement of the virtual environment?s performance to help identify critical bottlenecks and areas to be enhanced.