Research On Web Cache And Prefetching Model Based On Access Path Mining

With the development of Internet, especially the mobile Internet, the Web users are increasing rapidly. But since the increasement of network bandwith and the server number is relatively lagged behind, it leads to the latency of the visit by Internet users. Because of the enormous cost of hardware investment, cache system turns to be the most efficient way to solve the latency problem.Cache replacement algorithm has a strong impact on the performance of cache system. On the basis of researching the classic cache replacement algorithm GDSF, this thesis proposes a novel replacement algorithm IWAP-GDSF which is combined with the prediction queue created by Web log mining to improve the hit radio of the algorithm.Firstly, this thesis studied the Web log mining algorithm. Then, by thinking of the locality of Web object, a new mining algorithm which is based on the classic WAP algorithm, named IWAP, is proposed to adapt the prefetch model of the replacement algorithm. After that, this thesis proposed a new prefetch model including the position and count of the predict object in the predict sequence which represents the probability of the object’s future visit to provide the basis for the decision by replacement algorithm. Meanwhile, by combining the predict model, this thesis proposed a new replacement algorithm named IWAP-GDSF. The new algorithm considers the frequency, the cost and the probability for future visit of the Web object to make the replacement desicion. At last, this thesis programed to simulate the users’ visit behavior, and conducted the experiment of IWAP-GDSF algorithm and GDSF algorithm. The experiment results show that IWAP-GDSF algorithm could improve the hit radio and the byte hit radio effectively under the condition of the certain cache space.