A policy-based message broker for event-driven services in B2B networks

Abstract

Today's global economy requires businesses to expand their processes beyond organizational domains to integrate business partners and collaborators. SOA through Web services is emerging as a framework for bridging heterogeneous business systems over the Internet. Unfortunately it is still limited by many of the constraints of the legacy applications: procedural interaction, data polling and strong coupling of applications. Event-driven systems like publish/subscribe on the other hand provide a flexible interaction pattern that does not require rigid transport or communication protocol. Robust B2B process integration requires a framework that supports enterprise and Internet scale data sharing, point-to-point, push interaction as well as publish/subscribe broadcast based interaction. In our thesis, we accomplish this through a framework the views a B2B network as a streaming database and uses declarative policies to describe data sharing based on an SOA publish/subscribe infrastructure. The key contribution of our thesis is to define a flexible policy approach for describing streaming data. A Palliative Severe Pain Management scenario was implemented to evaluate our framework against BPI frameworks.