Design And Implementation Of Distributed Proxy System Based On Autonomic Computing

Nowadays the World Wide Web has in-depth to many aspects of human andsocial development, and it has become an integral part. With the development ofnetwork technology, all kinds of network services have appeared, however, becauseof large-scale network applications and services，a surge of network traffic hascaused network congestion, network latency and so on. What’s more, the percentageof network traffic produced by HTTP requests is increasing steadily, and the load ofWeb servers is also growing, while people has much higher and higher expectationfor the response speed of HTTP requests. And proxy technology is an importantmeasure, which can reduce Web server access load and improve the response speedof HTTP requests. In order to meet the massive traffic, the distributed proxytechnology appeared, and it can provide better network service. At the same time, ituses load balance technology and cache technology to improve the overallperformance of the distributed proxy system. In order to maximize the serviceability of cache servers, the method of evaluating content heat based on sub-domainpolymerization is proposed, and in order to manage the distributed proxy systemflexibly and automatically, the management mechanism of distributed proxy systembased on autonomic computing is proposed, and autonomic computing andreinforcement learning are introduced to this mechanism.Firstly, the theory and technology related to the topic is introduced, andunderstand the structure, components and function of the MAPE-K autonomic cyclemodel. And understand the several popular cache replacement algorithms and itsadvantages and disadvantages deeply. And understand the principle and applicationsof Q-learning by the way of example. Secondly, the design of distributed proxysystem based on autonomic computing is introduced, and the framework ofdistributed proxy system is proposed, which consists of load balance module, proxymodule and cache module. And the model of autonomic distributed proxy system isproposed, which combines autonomic computing with distributed proxy system. And in order to move the heated URL and reduce the load of cache node, themethod of evaluating content heat based on sub-domain polymerization is proposed,it not only thinks about the hit numbers of cache object, but also consider the overallcharacteristic of cache object. Then, the implementation of distributed proxy systembased on autonomic computing is introduced, and introduce the self-decisionmodule based on Q-learning, what’s more, the model and algorithm of Q-learningbased on BP neural network. Finally, according to analyze the experimental resultsof the framework of the self-decision based on Q-learning and content heatevaluation, it indicates that distributed proxy system based on Q-learningmanagement mechanism can effectively and automatically manage system, andimprove the overall performance of the system.