A RISC approach to GRID

Current GRID technology provides users/programmers with extended and comprehensive middleware tools covering all the basic aspects a programmer must deal with when writing GRID aware applications. Programmers of GRID aware applications use such middleware systems to gather the needed resources, stage code and data to the selected computing resources, schedule computations on them and so on. Overall, a huge programming effort is required in order to come to a working, efficient GRID aware application code. Here we propose a different approach. By recognizing that most GRID applications share a common parallel/distributed structure, we propose to use application managers that take care of all the details involved in the implementation of well known GRID aware application schemas. Such application managers use the functionalities of a RISC GRID core run time system (RGC). The RGC may be built on top of existing GRID middleware. Programmers of GRID aware applications, therefore, do not directly use GRID middleware. Rather, they structure the application using the available application managers. The application managers, in turn, invoke the basic GRID services provided by RGC to accomplish both functional and performance user requirements.