Title: Communication issues within high performance computing grids

Authors: Robert D. McLeod, Sajid Hussain, Sheng Huang, Marek Laskowski

Addresses: Department of Electrical and Computer Engineering, University of Manitoba, Winnipeg, Manitoba, R3T 5V6, Canada. ' Jodrey School of Computer Science, Acadia University, Wolfville, Nova Scotia, B4P 2R6, Canada. ' Department of Electrical and Computer Engineering, University of Manitoba, Winnipeg, Manitoba, R3T 5V6, Canada. ' Department of Electrical and Computer Engineering, University of Manitoba, Winnipeg, Manitoba, R3T 5V6, Canada

Abstract: This paper presents several ideas pertaining to desirable properties associated with protocols for moving large amounts of data over a grid of high performance computers. The protocol enhancements are discussed in terms of their scalability and identifying sources of potential protocol redundancy. In the context of this paper it is scalability as it relates to the transport of increasingly larger amounts of data over grid computing networks. Specifically we address efficacy in moving large data sets between nodes on a grid interconnected over a Wide Area Network (WAN) in terms of increasing the basic packet size and better suited to transport protocols. In addition we introduce a means of better evaluating error detection models as functions of increased packet size.

Keywords: transport layer; data transfer; rate based protocols; error detection models; high performance computing; grid computing; wide area network; WANs; packet size.

DOI: 10.1504/IJHPCN.2006.013480

International Journal of High Performance Computing and Networking, 2006 Vol.4 No.5/6, pp.248 - 255

Published online: 01 May 2007 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article