Research And Implementation Of Efficient Message Delivery Mechanisms With Reliablility Assurance In Publish/Subscrible System

With the rapid development of distributed technologies and cloud computing, we put forward higher requirements on the size and dynamic topology of the distributed system. Distributed systems are required that all their participants are able to have good dynamic and loosely coupled properties. The Publish/Subscribe systems are adapted to this demand compared with the traditional communication model (such as shared space, RPC/RMI, message delivery, etc.) in many aspects such as the characteristics of anonymous, asynchronous and many to many communication, supporting completely decoupled over time,space and control, thus it has been arising a great concern.Many cloud systems have added the Publish/Subscribe service. In this way, we just need to send messages to the cloud, then, can rely on cloud services to match and disperse the messages to each subscriber. As we know the data size of the cloud system is very large, for some large enterprise, hundreds of millions of messages are delivered per day, therefore, how to enhance the performance of messages delivery, and how to ensure the reliability of the delivery of the messages, is a very urgent and significant issue.Many research about enhance messages delivery performance have been studied at home and abroad, but they are more focused on the study of the reliability and Routing algorithms over the transmission network. There are also some studies on the messages delivery in real-time, such as a model-based priority queue algorithm and the shortest path tree algorithm, but they cannot guarantee the reliability of message delivery, and are also lack of attention on system terminal (publishers/message proxy servers/subscribers).This paper focuses on how to enhance the performance of message delivery with reliability assurance in Publish/Subscribe system. A lot of research on how to improve the performance of message delivery with reliability assurance has been done in the paper and the choice of pull/push message pattern, the consuming rate and the server side’s load balancing strategy are considered to be three key factors. Current message-oriented middleware, publish/subscribe or message queue, seldom concern about all these aspects. We integrate push message pattern and consumers cluster together with reliability assurance and meanwhile we offer a more fine-grained load balancing strategy based on message amounts. Finally all the mechanisms above are realized over a mature Publish/Subscribe system:Hedwig. Experiment has been made on PlanetLab, and the results show that it has better performance over messages delivery compared to the native Publish/Subscribe system.