Chapter

Grid Computing: Software Environments and Tools

pp 205-229

Programming, Composing, Deploying for the Grid

  • Laurent BaduelAffiliated withOASIS - Joint Project CNRS / INRIA / University of Nice Sophia - Antipolis, INRIA 2004
  • , Françoise BaudeAffiliated withOASIS - Joint Project CNRS / INRIA / University of Nice Sophia - Antipolis, INRIA 2004
  • , Denis CaromelAffiliated withOASIS - Joint Project CNRS / INRIA / University of Nice Sophia - Antipolis, INRIA 2004
  • , Arnaud ContesAffiliated withOASIS - Joint Project CNRS / INRIA / University of Nice Sophia - Antipolis, INRIA 2004
  • , Fabrice HuetAffiliated withOASIS - Joint Project CNRS / INRIA / University of Nice Sophia - Antipolis, INRIA 2004
  • , Matthieu MorelAffiliated withOASIS - Joint Project CNRS / INRIA / University of Nice Sophia - Antipolis, INRIA 2004
  • , Romain QuiliciAffiliated withOASIS - Joint Project CNRS / INRIA / University of Nice Sophia - Antipolis, INRIA 2004

* Final gross prices may vary according to local VAT.

Get Access

9.5 Conclusions and Perspectives

In summary, the essence of our proposition, presented in this chapter, is as follows: a distributed object-oriented programming model, smoothly extended to get a component-based programming model (in the form of a 100% Java library); moreover this model is “grid-aware” in the sense that it incorporates from the very beginning adequate mechanisms in order to further help in the deployment and runtime phases on all possible kind of infrastructures, notably secure grid systems. This programming framework is intended to be used for large scale grid applications. For instance, we have succeeded to apply it for a numerical simulation of electromagnetic waves propagation, a non embarrassingly parallel application [21], featuring visualization and monitor- ing capabilities for the user. To date, this simulation has successfully been deployed on various infrastructures, ranging from interconnected clusters, to an intranet grid composed of approxi- matively 300 desktop machines. Performances compete with a previous existing version of the application, written in Fortran MPI. The proposed object-oriented approach is more generic and features reusability (the component-oriented version is under development, which may further add dynamicity to the application), and the deployment is very flexible.