High performance parametric modeling with Nimrod/G: Killer application for the global grid?

The evolution of a particular tool, Nimrod, from a local computing environment to the global computational grid is discussed. The various services of Globus and how these were applied in building a grid aware application are described. The algorithm used is both simple and adaptive to changes in the workload distribution on the grid and incorporates user requirements as well as system ones. Results show that it is possible to build an application which takes account of the highly dynamic and unpredictable nature of the grid