GridDaen: A Data Grid Engine1

The volume of datasets in scientific computing domain is increasing explosively and data is becoming the center of scientific computing. Data Grid is an emerging technology to provide uniform access and management of the large scale distributed scientific datasets. In this paper, the GridDaen system, a new Data Grid middleware, which provides uniform APIs and GUI to access and manage resources, is presented in detail. GridDaen utilizes a three-level naming scheme that shields users from low-level resource discovery and provide global uniform view for users. Coordinated DRB servers and multiple -layer distributed metadata servers are adopted in GridDaen to provide uniform secure access and management of heterogeneous distributed resources across multiple administrative domains. Role-based multi-level access control policy is used in the system and caching and replication mechanisms are utilized to improve the performance of the system. The virtual dataset functions are also supported in GridDaen. GridDaen can achieve good scalability, reliability and can be flexibly deployed and configured according to the application demands.