A Graph Database Approach for Efficient and Scalable Management of Simulations

Relational database technology has enabled easy and efficient management of data in a variety of applications. This technology, however, has been used for computer simulations to a limited extent. In this context, users may need to analyze the results of multiple simulation runs, which typically capture interactions between various entities in the simulated world. To these users, storing such simulation data only in the form of relational tables may seem counter-intuitive or challenging, particularly if they are not familiar with database normalization theory. Furthermore, some complex queries, including those that examine the cumulative effect of an action (e.g., disease propagation after an initial outbreak), can neither be easily expressed nor efficiently executed on traditional database systems. We propose a new database approach that aims to achieve convenient and highly efficient storage and querying of simulation data by adopting a graph data model. We also discuss new challenges that arise in this research, with a focus on language design, coordination of simulations, data storage, and query processing.