Research On Replication Strategies In Educational Resources Sharing Grid

With the development of educational informatization, the educational resources are increasing more and more rapidly, which not only helps education service a lot, but also brings new challenge for construction and management of education resource. Nowadays, the construction and management of educational resources in our country are lack of uniform supervision and standard for quality of resources. It is important to achieve resources sharing and make the seperate and out-of-order resources to be conformed.Some experts suggest educational resources sharing grid as the solution for problems mentioned above.Grid computing is a novel and emerging paradigm, whose goal is providing virtual organisations of geographically, distributed users with computaional and storage resource. Educational resources sharing grid is a new type of data grid.In data grid, how to make all the users from all over the world to access the data set effectively is a big challange.Replication, creating geographically disparate identical copies of data, is regarded as one of the major optimization techniques for resolution. Probable replication strategy can reduce access costs and bandwith consumption.It also helps to improve system perfomance and fault tolarance.By consulting a large number of relevant documents and analysis on combining my own theory, I have found that current existing replication strategies are fit for general data grid, and not for educational resources sharing grid.So, on the basis of studying current data grid replication strategies, this thesis explores and designes a new file content-based data grid replication strategy for educational resources sharing grid.This replication strategy includes grid structure, replica creating, replica replacement and replica selecting.This replication strategy adapts P2P-based region-centered grid structure.The grid is divided into several regions, for which the bandwith between regions is small and bandwith in region is large. There is a center site in every region which stores file information of its region.Replica creating and replica replacement are based on users'interestingness in files.The replication strategy introduces knowledge to express content of file and uses user access history to predict the hot content in the future. For the files that are not accessed, this strategy calculates the content similarity of them with the accessed files.More similar the content, closer user interestingness. Further more, for the purpose of educational resources optimization, this thesis designs a user judgement function, which shows the tendency of users'attitude for files.This replication strategy adapts auction to select proble replica, which costs the shortest transmission time.To check the repliction strategy's performance, this thisis uses OptorSim.The result from simulation in OptorSim shows that this strategy can outperform other optimization techniques in educational resources sharing grid.