Research On Web Cache Technology

With the rapid growth of the size of information and users on the internet, it becomes more and more urgent and difficult to effectively reduce the user access latency time, and improve network service quality and network bandwidth utilities. Proxy caching technology of web is the most effective way to overcome these problems. This thesis analyses and researches in depth on the cache replacement strategy, cache consistency technology and prefetching technology. On the basis of previous analysis, this thesis focuses on research and improvement of prefetching technology.First of all, this thesis introduces performance problems Internet confronts and solutions to them, and then elaborate the basic concept and theory of proxy caching technology of web, prefetching techniques and the cache replacement strategy. Next, this thesis also classifies related research results, points out the things need to improve, which lays the foundation for further research.Secondly, based on analysis of trajectory documents, this thesis propose a new strategy to improve LRU, which not only considering the average quote time, the recent time lapse, document type and size, and file age, but also considering the dynamically changing features of access rate.Thirdly, at the cache consistency strategy, this thesis improves volume lease updating mechanism and TTL mechanism. The new TTL mechanism overcome the shortcomings of TTL mechanism by the return of a series of versions of requested web pages and improve the calculation of time lapse and allow customers to specify the old degree that the request have to meet. Volume lease mechanism is fully taken into account the different changes of the data objects and network resources, and specifies different expiry dates respectively. Volume lease mechanism maintains the consistency of object more accurately.Finally, prefetching threshold algorithm is improved. Based on comprehensive consideration of network performance and Web server capacity, this thesis proposes threshold factors. The improved threshold algorithm dynamicly predict prefetch threshold of web page prefetching algorithm base on user access path analysis. Experiment illustrates that after dynamically prefetching a threshold, performance of the new algorithm has been greatly improved.