Research On Some Key Technologies Of Mass Data Management Based On Data Grid

With the quick development of the information technology, data explosion has become a challenging problem, which leads to that mass data management technology becomes a new focus and hotspot in the research area. With the Digitized Virtual Human as background, coupled with Hi-Tech Research and Development Program of China (863)"Construction of Digitized Virtual Chinese Human Data Sets and Mass Database", this thesis put emphasis on mass data management technologies. The main work and contributions are as following:1. Mass data management method based on Data GridData Grid is one of the branches of the Grid technology and it puts more emphasis on data store and management. Based on current data management technologies, Data Grid aims to integrate all kinds of resources in the network, such as storage resources, data resources, computing resources, etc., and establish a unified architecture for data access, store, transfer and management, which offers a new way for mass data management.In this dissertation, the Data Grid idea was introduced into the Digitized Virtual Human research. Taking the characteristics of Digitized Human datasets into consideration, some solutions for key technologies of mass data management were presented, such as storage resources allocation, replica management, high speed data transfer and quick data access. Furthermore, a Data Grid-based Digitized Virtual Human mass data management prototype system was established. The system integrated distributed heterogeneous resources into a whole to form a virtual store environment, and provided applications a unified access interface, which realized data share in a great deal.2. Research on Data Grid metadata modelMetadata is one of the most important components of the data management. Most existing researches on Grid metadata pay attention to those data in file systems, while there is little work on the data in database systems. Consequently, a Grid database metadata model was put forward in this dissertation. Through a relational-hierarchy blended model, the metadata model realized a unified management for data in both file systems and database systems. All related information of the required data may be obtained through one direct location, which improved the inquiry efficiency greatly.3. Common structure for Digitized Virtual Human application platformA common structure for Digitized Virtual Human application platform was proposed coupled with the real application demands of the Digital Human. Combining mass data management technology, metadata model and middleware technology together, a data...