Research On Data Grid Middleware For Relational Data Sharing

Data resource is one of most valuable resources in the Information Age. Scientific data are the crucial foundation for emerging e-Science activities. Relational data are usually processed with serious design and therefore are of more importance. As such, the significance of sharing massive relational data among a huge number of users is self-evident. However, because of practical difficulties on resource discovery, heterogeneity, security, and so on, current users can not access those data from a relational database management system (RDBMS) directly through the Internet or Web yet. The existing technologies can not provide necessary support for sharing of relational data in a wide range. In particular, many such requirements come from research communities.In this dissertation, the author has managed to make use of the state-of-the-art data grid middleware technology to solve the issue of relational data sharing as above. In addition to those factors when considering an ordinary issue of resource sharing, the issue here has four characteristics or constraints as follows: users need to read data only, no data modification or management involved; those data with long-term value and shared by many people are usually not changed frequently, especially in terms of scientific data; different users have different rights on access data and the system must undertake permission management with a large range of users; the main method of data sharing is not to get a piece of data as the result of a query, but to copy or download a data resource systematically, demanding a capacity of high efficiency massive data transfer.According to the objective defined above, the following three key topics are studied in this dissertation: how to find data, how to get data, and how to grant permissions on data. The topic "how to find data" means that how a user could find the table containing data he/she wants from a large number of relational databases. The topic "how to get data" means that how a user could access comprehensible and ready-to-use data in a standard and handy fashion. The topic "how to grant permission on data" means that how a huge number of cross-organization users could acquire reasonable permissions to share data through a simple, convenient and flexible mechanism.Here are the major contributions of this dissertation:...