Analysis And Improvement Of Grid Large File Transfer Based On GSI And GridFTP

In modern scientific research and application areas, the large amounts of data are important resources, and the data may reach the level of TB or even PB. The data stored in the WAN multiple sites require to meet the needs of researchers to share. In recent years, the Data Grid provides an effective way to solve the above problem. And the Globus project is one of the most influential Data Grid research projects.On the basis of European-China Grid Internetworking project of Beijing University of Posts and Telecommunications (BUPT), this paper focused on analysis and improvement of large file transfer based on the Grid Security Infrastructure GSI and GridFTP protocol.Firstly, the physical structure of this system consists of one server node and two client nodes in the project. And we configured the Globus Toolkit's GSI and GridFTP protocol on the nodes to meet the use of GridFTP transfer required, and implemented the GridFTP transfer between Linux platforms. Secondly, we tested the basic functions of GridFTP transfer, including the certificate authentication, transfer between the client and server, third-party transfer between servers and parallel transfer, etc. And then we analyzed the traffic feature of the large file transfer in the Grid. We developed a new algorithm to improve the existing congestion control mechanism, and integrated the new algorithm into the GridFTP protocal. Simulation and validation through improved transmission performance of GridFTP was made to validate the grid environment for the large amount of data transmission.Finally, by comparing the GridFTP transfer with FTP transfer, we proved that the improved GridFTP protocol was more suitable for large file transfer and had a better performance. In this paper, the research and development has a positive significance to European-China Grid Internetworking project of BUPT. Analysis and improvement of large file transfer is useful for the research of the GSI security mechanism and GridFTP. Furthermore, comparison and analysis of the performance of the improved GridFTP transfer with FTP transfer provides the foundation for the research of massive date transfer on the project.