Service Oriented Grid Architecture for Geosciences Community.

Grid computing has many attributes that provide benefits to research communities. These include the ability to do existing research faster; enable collaborative environments; reduce costs and provide a pool of resources that creates an ability to conduct greater research. These have been identified by the Geosciences community as of significant interest and a plausible solution to some of the community's current linkage and development problems. The Geosciences community for sometime now have often been restricted to "small" science, easy access to greater computational and data resources that would allow larger experiments to be conducted. The Grid poses a solution to this problem by providing researchers simple uniform access to a myriad of resources via a single-sign-on policy. The Australian Partnership for Advanced Computing (APAC) Geosciences Grid project together with Solid Earth & Environment Grid (SEE Grid), aims at providing Computational Geoscientists with uniform access to some of the more common Geosciences-based computational codes on large computational and data resources available via the grid. To achieve this goal, an interoperable approach has been taken and web/grid services where used to establish the links to such resources. The Grid's service oriented architecture enables the programming of specific self-contained services that simplified the access to computational and data resources for the user. Through the use of the Globus Toolkit 4, we have developed grid services that assist the common Geoscientist user to collaborate and conduct large-scale experiments including parameter sweeps. The deployment of these interoperable grid services have allowed us to develop multiple client programs in varying languages in an effort to give researchers the choice of tools to use to interact and conduct their research. We have also used existing client tools to connect to the. Such interfaces have been deployed to several Geoscientists and into the pmd*CRC.