Parallel Architectures, Algorithms and Programming, International Symposium on (2011)
Dec. 9, 2011 to Dec. 11, 2011
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/PAAP.2011.71
The biggest problem to be solved in grid computing is data transmission and communication efficiency and stability. For computing grid composed of small local area network(LAN), this problem can be solved by properly building LAN with the use of IPv4 network or high-speed computing network such as Infiniband. However, when the computing resources are distributed in different physical regions, data transmission and communication will be affected by various factors such as time period and load. On the one hand, the stability of transmission cannot be guaranteed; On the other hand, the speed of cross-regional data transmission is not optimistic. As the next generation of Internet protocol, IPv6 is at a transitional stage, and the load is far less than the IPv4-based network. So, it has a better stability and higher bandwidth. With the help of IPv6 network, cross-regional computing grids can be constructed to connect the computing resources in different physical regions in order to deal with some complex computing problems. This paper studies and tests the feasibility of transplanting MPI application (which is more commonly used in distributed computing) into IPv6 network for grid computing, and presents its implementation process in detail.
grid computing; IPv6; MPI
Y. Zou, J. Chen, Z. Liu and Q. Wu, "Enabling Grid Computing over IPv6 within a Campus Network," Parallel Architectures, Algorithms and Programming, International Symposium on(PAAP), Tianjin, China, 2011, pp. 285-288.