PERFORMANCE EVALUATION OF GRID FTP PROTOCOL

Abstract

Grids have become a critical asset in large-scale scientific and engineering research. Accessing distributed data is normally as important as accessing distributed computational resources. These resources are shared over the Grid. This sharing is, necessarily, highly controlled, with resource providers and consumers defining clearly and carefully just what is shared, who is allowed to share, condition under which sharing occurs. Most of the advanced applications in Grid environment require access to remote data. Therefore while dealing with remote file transfer, many aspects, such as which protocol is supported during the transfer, how the file transfers are organized, which mechanisms are used to instantiate a file transfer etc, must be considered. GridFTP is a high-performance, secure, reliable data transfer protocol optimized for high-bandwidth wide-area networks. The existing data transfer protocol such as FTP is not suitable for data transfer on high-speed networks, as it does not utilize the bandwidth efficiently. Further, it does not have any support for (a) Third Party Controlled Transfer of data & (b) Parallel Transfer of data. Hence, a need was felt to develop an Application level protocol to meet the standard needs and features required by the modern day Grid applications. This Dissertation presents the extensions; including their implementation, to the standard FTP protocol to make it suitable for the Grid environment. These extensions enable the standard FTP protocol to provide features that were previously not available such as transfer of data between two servers, in which the data transfer between two machines is controlled by a third party. Another feature that enables transfer of data over multiple data connections, i.e., there are more than one data connection between the client and the server, which results in efficient utilization of the bandwidth. The security requirements for data transfer on the grid are also discussed