Research And Implementation On Collaborative Technologies Across VOs In Data Grid

Nowadays, more and more applications adopted Data Grid technology as their data management infrastructure. With this trend, the diversities in Data Grid become more and more notable, including the diversity of data resources, virtual organizations (VOs) and so on. While the branches of knowledge are crossing each other and inter-disciplines are developing rapidly, the Data Grid needs to collaborate between VOs and share distributed resources efficiently and securely.Considering the diversities in Data Grid, in this paper, we analyze the imperative requirements of collaboration between VOs, and meet the challenges by the following two approaches:Firstly, a Grid Data Transfer System - GridPard - is designed and implemented, which provides advanced characteristics, such as parallelized and stripping transfer, common and indirect third-party transfer, transfer-with-route mode, and so on. With these characteristics, distributed, heterogeneous, and massive data can be transferred between arbitrary VOs, while the efficiency, feasibility, stability, reliability and security of the system can be guaranteed. Experiments show that our system can obtain a magnificent performance improvement against traditional systems.Secondly, a new mechanism, called Negotiation-based Dynamic Role Transition in Data Access across Multi-VOs, is proposed. It enables the secure data sharing between VOs. This mechanism sets up role-mapping relationships between VOs, and allows users to access data resources in the other VO by dynamic role transition. It keeps one VO's role-hierarchy transparent to the other, and adapts to the roles' adding and deleting. Meanwhile, our experiment shows that the overhead of the role transition is relatively low, and the mechanism is correct and effective.The Grid Data Transfer System guarantees high-quality data transfer in data sharing, while the mechanism of data access control across multi-VOs guarantees high-security data access between VOs in data sharing. Experiments show that, the collaboration between VOs becomes more secure and efficient, in the diverse Data Grid environment.