Research Of Cache Management In "Papershare"

With the rapid development of Internet, the independent Web Cache technology hasbeen widely appreciated and developed. The Relational database cannot meet the needs ofscalability, fast data backup and recovery and the ease of use. The technology ofindependent Web caching has been widely appreciated. We write data to cache inkey-value pairs. Then we can read the data directly from the memory. This method canimprove the system’s service capacity effectively.The cache management systems of the current Internet applications face the problemof the cache pollution caused by data diversity. Considering the specific issues of the“PaperShare” system, we propose a new technology that store different types of data intodifferent data structures in memory rather than putting them into the single Hash table.This can solve the problem of the interference of different types of data. Thereby improvethe performance of the “PaperShare” system. We chose a distributed algorithm and dosome developments to improve the distributing balance of the algorithm. And then wedevelop the scalability of our cache system using the algorithm. This can solve theproblem of the lack of memory in one computer.The results show that the method of storing data according to its content reduced theaverage response time of the “PaperShare” system by60percent. The read and writeperformance of the “PaperShare” system’s cache tripled on account of the distributedexpanding. The improvements of the distributed algorithm achieve good results whichmaking the data distributed evenly across the nodes of the distributed cache system.