A Standard Interface For Grid-Powered Parametric Optimization

Today’s most powerful computers are mainly based on cluster architectures, however, small and medium enterprises (SMEs) often cannot afford the investment necessary to build and manage a cluster. A dynamic solution based on grid computing allows generating dynamic clusters from already in house, nondedicated machines, however parallel programming requires a specific know-how that is quite seldom available in SMEs. Among engineering practitioners, one of the most commonly used tool is a spreadsheet such as Microsoft Excel. At present, however, many of the problems confronted with have to be oversimplified to cope with the fact that Excel can be run only on a single Win32 box. In this paper we describe Excelerator, a tool capable of i) extracting optimization problems from Excel worksheets, ii) deploying the computational load onto a small sized computational grid such as the one that can be found inside a small-medium engineering firm and finally iii) presenting the results back inside an Excel worksheet.