Sub-2-Sub: Self-Organizing Content-Based Publish and Subscribe for Dynamic and Large Scale Collaborative Networks
Abstract
In this paper, we address the problem of constructing scalable content-based publish/subscribe systems. Publish/subscribe systems are asynchronous event-notification systems in which a published event is forwarded to exactly those nodes that have previously subscribed for that event. Subscriptions can range from a simple specification of merely the type of an event to a specification of the value ranges that an event's attributes can have. Notably the latter poses potential scalability problems. Structured peer-to-peer systems can provide scalable solutions to publish/subscribe systems with simple subscription patterns. For complex subscription types their applicability is less obvious. In this paper, we present Sub-2-Sub, a collaborative self-organizing publish/subscribe system deploying an unstructured overlay network. Sub-2-Sub relies on an epidemic-based algorithm in which peers continuously exchange subscription information to get clustered to similar peers. In contrast to many existing approaches, Sub-2-Sub supports both value-based and interval-based subscriptions. Simulations of Sub-2-Sub on synthetic and reusable workloads convey its good properties in terms of routing efficiency, fairness, accuracy and efficiency.