基于自决策的分布式代理缓存技术分析-analysis of distributed proxy caching technology based on self - decision.docxVIP

AbstractWith the development of IT, the Internet has in-depth to all aspects of people’s life. At the same time, various web services began to rapid growth. However, the huge network traffic has caused network congestion between users, and results in the decrease of network quality of service and user experience. Internet Service Provider needs a kind of technology to reduce network traffic load and service delay time. Since the HTTP traffic has a high percentage of total network traffic, HTTP proxy caching techniques can not only effectively solve above problems, but also reduces network traffic load. In order to meet the massive traffic, the distributed proxy cache system appeared, and the performance of this system depends on the cache management technique. For achieving high-efficiency and automated cache management, this paper proposes the mechanism of proxy cache management based on self-decision, and self-awareness and self-decision are introduced to this mechanism.At first, self-management model and proxy cache management methods based on self-decision are introduced, and built the self-management framework according to the idea of feedback loop. Then, the mechanism of distributed proxy cache management based on self-decision is proposed. In order to analyze the monitor items and identifying the state, the weighted average method based on correlation coefficient is proposed. In order to uniformly distribute URL space between cache nodes, the mechanism of hash ring with virtual node is proposed. In order to select the cache nodes need to be adjusted, sequence interval constraint classification method is proposed. In order to calculate how to adjust cache node, the state adjust strategy based on historical states sequence and grey forecasting model is proposed. In order to move the heated URL and reduce the load of cache node, the method of evaluating content heat based on domain polymerization is proposed. Next, depending on self-management framework of proxy