Architecture, performance, and scalability of a real-time global positioning system data grid

We describe the architecture of our real time SensorGrid system, which supports the real-time message routing and processing of Global Positioning System data. The current system processes 1 Hz position data from 85 stations in Southern and Central California. Our system is built on a computer network-enabled, topic-based publish/subscribe system and is extensible to other data sources. In order to determine the performance and the upper limits of scalability, we have conducted and present here a set of systematic test evaluations of our implementation. Through these tests, we are able to show that performance does not degrade over time, that the system will handle up to 1000 simultaneous data providers or data consumers with a single message broker, and that multiple message brokers can be linked to provide scalability beyond 1000 providers or consumers.