Subscription tree pruning: A structure-independent routing optimization for general-purpose publish/subscribe systems.
Citation
Export citationBittner, S. & Hinze, A. (2006). Subscription tree pruning: A structure-independentrouting optimization for general-purposepublish/subscribe systems. (Working paper series. University of Waikato, Department of Computer Science. No. 01/2006). Hamilton, New Zealand: University of Waikato.
Permanent Research Commons link: https://hdl.handle.net/10289/52
Abstract
A main challenge in distributed publish/subscribe systems is the efficientand scalable routing of incoming information (event messages). For largescalepublish/subscribe services, subscription forwarding has been established asa prevalent routing scheme. It reduces the network traffic for event routing dueto selectively forwarding event messages to relevant parts of the network only.To further improve event routing, publish/subscribe systems apply routing optimizations.So far, optimizations for general-purpose publish/subscribe systemsare still missing.In this paper, we present the architecture, realization, and evaluation of our prototypeof a large-scale publish/subscribe service applying a novel routing optimization,subscription tree pruning. We also show a comparison of five existingrouting optimizations in respect to six important characteristic parameters affectingthe suitability of these approaches in practice (including space usage, timeefficiency (throughput), and network load). This comparative analysis clearlydemonstrates the advantages of subscription pruning over other routing optimizations.In our practical experiments, we then investigate the behavior of our prototyperegarding all quantitatively measurable parameters from our previouslytheoretically analyzed ones. Our evaluation of subscription pruning in this paperis more extensive than previous analyses of any routing optimizations for publish/subscribe systems, which focus on selected parameters only.
Date
2006-01-01Type
Report No.
01/2006
Publisher
Department of Computer Science
Collections
- 2006 Working Papers [10]