Parallel processing on heterogeneous networks for GIS applications

The use of network based parallel computing is gaining increasing popularity for different reasons. Its exploitation depends on the availability of simple but effective methodologies to parallelize applications, and the availability of portable and efficient communication libraries to develop parallel programs. These two items are necessary to obtain advantages for what performance is concerned, and to ensure software portability and reusability. In this paper we present our experience in parallelizing, in a systematic way, a class of GIS (Geographic Information Systems) applications. We also discuss the use of two well known communication libraries (PVM and Linda). Performances results are also reported.