Design And Implementation Of A Lightweight Grid-Enabled MPI-Like Parallel Programming Framework

The emergence and rapid development of Grid computing have changed the traditional computing paradigms a lot. Grids enable such applications as large-scale computation, distributed data analysis to get more computational resources and more computational capabilities to an unprecedented extent by integrating various kinds of geographically distributed resources and providing uniform and transparent access to those heterogeneous resources.Application development for Grids has been benefiting from various programming tools and development environments that facilitate the convenience of the building process to some extent. However, most such implementations have some disadvantages and limitations. Some programming models that they provide are overly specialized high-level approaches that are not likely to be applied elsewhere. Worse than that, some implementations come along with very complicated development and execution processes that require laborious configuration and deployment of certain components. Being too dependent on the specific Grid middleware can also be a hindrance to scalability and portability. The lack of support for coupled and coordinated use of Grid services is another big problem in today's service-oriented Grids.