As discussed above, we envision the middle tier of the network computing architecture providing the interface to high-performance computing capabilities. In the CORBA based strawman that we have presented, this means that high-performance computing components must be integrated into the CORBA architecture.
CORBA is defined in terms of a set of facilities, where each facility defines an established, standardized high-level service. Facilities are split those that are required by most applications (called horizontal facilities) and those facilities that are defined to promote interoperability within specific application domains. As CORBA evolves, it is expected that some vertical facilities will migrate to become horizontal facilities.
We believe that high-performance computing can be integrated into the CORBA model by creating a new facility which defines how CORBA objects interact with one another in a high-performance environment. CORBA currently supports only relatively simple computing models, including the embarrassingly parallel activities of transaction processing or dataflow. High-performance commodity computing therefore would fill a gap by providing CORBA's high-performance computing facility.
This new facility allows us to define a commercialization strategy for high-performance computing technologies. Specifically, academia and industry should experiment with high-performance commodity computing as a general framework for providing high-performance CORBA services. Then, one or more industry-led groups should propose high-performance commodity computing specifications, following a process similar to the MPI or HPF forum activities. Such specifications could include another CORBA facility, namely, that involved in user interfaces to (scientific) computers. This facility could comprise interfaces necessary for performance tools and resource managers, file systems, compilation, debugging, and visualization.
We note that we focus here on the use of CORBA, analogies exist in the Java and COM object models. In particular, in Section 1.6, we discuss how wrappers might be used to provide a Java framework for high-performance computing.