Adaptive load balancing and fault tolerance in push-based information delivery middleware service

Push-based personalized information delivery is a special kind of event-driven application, which must monitor or react to changes of information sources or users interests to provide user-centered real-time services. In this paper, an enhanced event-driven middleware framework of Quality of Service (QoS) is proposed to support high reliable and available timely information push-delivery, in which adaptive loading balance strategy that combines static and dynamic load distribution algorithms together to ensure the system works in a reasonable workload level. As for the fault tolerance, a super-demon with adaptable time-piece is adopted for failure detection, node replication and recovery. Large amount simulations showed the adaptable QoS strategies and algorithms that we proposed are effective, efficient for event-driven distributed systems.