Understanding the characteristics of information flow in large scale real-time distributed virtual simulations (RT-DVS) is important for the development of network services that are able to meet the robust needs of this simulation environment. Being able to quantify these characteristics enables network performance estimations to support the growing demand for use of Internet/Web-based services in large scale RT-DVS applications and interest in simulation interoperability with Communications, Computers, Intelligence, Surveillance, and Reconnaissance (C4ISR) systems. This simulation environment requires open network communications protocols that can accommodate the efficient distribution of large amounts of data to many users, which in turn implies the use of many-to-many multicast network services that are not readily available as an open Internet service today. This paper describes results from the study of network information flow in three simulation experiments and presents a supporting analytical model for use in predicting performance of an early implementation of the Extensible Modeling and Simulation Framework Overlay Multicast Protocol (XOM). XOM is designed to support many-to-many multicast for efficient exchange of real-time information among many users in real-time simulations over the Internet. We summarize the architecture and key design considerations of XOM that result from these live simulations studies and our measurements of network traffic and the resulting analytical model.