Scalable Dynamic Synthetic Environments using a Next Generation Game Architecture

For some years now, the M&S community is transitioning from the traditional large, monolithic stove pipe based systems to Modelling and Simulation as a Service (MSaaS). By running systems on virtual machines in the cloud, their accessibility and reusability is improved, and the deployment and configuration of hardware and software has become less cumbersome and time-consuming. The question remains, however, whether MSaaS in its current form is really the next-generation simulation architecture that will be able to cope with the ever increasing demands in scale, level of detail and accuracy of military simulation systems.
A current trend within the commercial game development community, to enable massive online worlds, is to develop microservice-based solutions and platforms (e.g. SpatialOS, Coherence, cloudgine, Amazon Lumberyard). These microservices promise to deliver on scalability (by off-loading large datasets and intensive computational work to the cloud and providing automatic load-balancing), correlation and fair-fight (e.g. by means of a shared world model), separation of concerns (SMEs can work on specific microservices, e.g. vehicle trafficability, physics-based destruction), centralized scenario configuration (versioning, management and monitoring), and more.
In this presentation, we discuss the application of a microservice platform to enable a dynamic synthetic environment, and see if it makes good on these promises. The experiment focused on a building destruction use case, with both a simple model-switching as well as a complex physics-based destruction approach implemented as a microservice running in a commercial cloud-based game platform. We examine challenges encountered while moving to the microservice paradigm and practical considerations when reusing a commercial game architecture for M&S, and give an outlook on the way ahead, providing our view on whether this is the right path towards a truly next-generation simulation architecture.