Research And Implementation Of Reliability Safeguard Mechanism In Push-Style Ws-Notification System

Though Internet undergoes a rapid development, there still exist critical constraints within "Request-Response" communication mode, which appears to be strong coupling between client and server. Moreover, it limits the extensibility and reuse of systems. Consequently, Publish/Subscribe technology emerges as a communication mode with low coupling, asynchronous and multiple to multiple. Its characteristics of asynchronous and multiple to multiple communication strengthen the extensibility and flexibility of distributed communication system, thus it’s intensively applied to a variety of SO A and cloud computing systems.As a widely adopted standard of Publish/Subscribe, Web Service Notification (WSN) sets up a series of Riles in Event-driven system of Web environment, including:the message exchange format, basic frame of Publish/Subscribe and mutual mode based on the agent between the communication substance in Publish/Subscribe system. At present, these is no further study on reliability on WSN Publish/Subscribe system in both academia and industry. On one hand, WSN doesn’t take into account the issues like recovery mechanism of Publish/Subscribe’s entities under failure as well as reliability of the service. On other hand, the main research related to the reliability focuses on the routers, it still remains to be open for the research such as how to establish a reliable Publish/Subscribe system.This paper will discuss a design of reliable mechanism for the service of subscribing and pushing based on WSN standard. It will lay emphasis on the reliability and self-recovery of subscribing service together with the stability and error-resend mechanism of service communication. We will start from the discussion of topological structure of distributed network, and then a subscribing-mutual backup relation will be achieved automatically between multiple agent nodes of the system, which recovers the subscribing service in case of troubles of nodes and sends out the notification messages in time through initiative push. Simultaneously, we have adopted a reliable mechanism which can ensure the reliability of Publish/Subscribe service including initiative push. The experiments show that through incorporating this design, the reliability of system has been improved while the efficiency has been preserved.