Complex Event Routing in Pub/Sub Systems Using Traffic Analysis Model

Bringing in complex events in distributed pub/sub systems enables users to observe complicated event correlations over time and space. However, the routing of complex events in pub/sub systems is still largely unexplored, especially lacking quantitative analysis and deep assessment. This paper focuses on the deployment of composite subscriptions, formulating the traffic cost by using the traffic analysis model, and proposes the traffic-driven deployment protocol (TDDP) in order to reduce network traffic in the system. TDDP is implemented on top of our pub/sub system OncePubSub and its performance is also evaluated. The results of the experimental evaluation show that adopting TDDP to deploy composite subscriptions can dramatically reduce the network messages than simply deploy them on the access brokers.