A self-managed self-optimized publish-subscribe system

Abstract

Publish/subscribe based communication systems have become very popular in recent years. Such systems are becoming larger and more complex, and thus require a smart management framework. An important challenge in this context is to efficiently disseminate the data flows sent from the publishers to the subscribers. To this aim, multicast dissemination is often used, requiring a smart mapping of data flows to multicast groups. Most existing publish/ subscribe systems use static configuration and thus do not efficiently handle dynamic changes in the publish/subscribe system. In this work, we present a self-managed and self-optimized publish/ subscribe system that efficiently adapts to changes in run time. A key element in the solution is a smart mapping algorithm that computes efficient routes for the data flows based on the current conditions in the system. Themapping algorithm takes into account various costs and constraints that are associated with the transition from one mapping to another during run time. The solution we present maintains the publish/subscribe system optimized while at the same time ensuring the stability of the system. We complement our work with a comprehensive simulation study in which we evaluate the suggested solution. The results clearly demonstrate the advantages of a dynamic self-optimized system over a static system. Copyright 2013 ACM.