A Study Of Cache Technique For Catalog Server

As the development of internet and software industry, using catalog tree in software resource management or to show the inner resource in website is more and more common. The conventional catalog tree is dynamically created by combining with the database, which not only reserves all the information of catalog tree in the database but also maintains itself by operating the database. But the efficiency of this kind of catalog tree is low when unfolding the sub-node and reading its information, especially under a huge data volume or visits. Existing catalog server optimizes the various operating of the catalog tree by catalog access protocol, but the performance is severely constrained when concurrent access is too high caused by the large amount of users.In the basis of the conventional catalog tree, we put forward a new dynamic system which can quickly and efficiently load catalog trees based on parallel processing and cache mechanism. The technologies we used building the frame of the system are popular right now. For example, Spring makes the frame more concise and Hibernate improves the unfold and modification performances of the catalog tree node.In the new system, a physical memory space, which has been divided into several physical memory blocks of different size, is applied to as a data buffer. The frequent-login catalog tree nodes are reserved in different blocks which are locked by mutual exclusion. As a result, many users can access or modify their own catalog trees at the same time. That is the parallel processing of reading catalog trees which adopt a multiple cache data replacement strategy to ensure the consistency and enhance the hit rate of the cache data by estimating Hash index to accelerate the searching speed. Compared with the conventional catalog tree, the capacity of this new system which combined with cache and multi-threaded mechanism has been greatly improved.