Research And Implementation Of Distributed Gridware Environment And Tools

The rapid development of network distributed computing, have brought newopportunities and challenges in improving software performance. Ideally, we can increase theinvestment in hardware to improve the software performance, or realize large-scaleapplication by integrating multiple software systems. However, the traditional software model,such as object-oriented, developed under the assumption of a single control flow, an objectcan contains shared state that makes parallel computing difficult to achieve, so it can’tdirectly transform into a distributed development model. In contrast, data flow model is verysuitable for distributed parallel computing, but because of the lack of common software modelelements such as incident response, processes services, non-deterministic, has not becomemainstream software model.In this context, a research team from South China University of Technology, leaded byProfessor Deyu Qi, raises a new calculation model, the Gridware Model, based on the ADL,this model is a "problem solving machine". The Gridware Model introduces new systemstructure component, including field and grid, and their basic operation, including field fusionand grid fusion. In the Gridware Model, a problem-solving process is divided to construct thefield, mutate the fusion device, construct and roll-out expand the Gridware, and criss-crossfuse the fusions.The Gridware Model is based on the data flow computational model. The GridwareSystem, based on the Gridware Model, is a software development framework with the naturalconcurrency, can be used for distributed development, can shorten development cycles ofmulti-threaded system, reduces software development costs, solves the software crisis,improves software productivity and quality. This paper studies the Gridware Model, sums upthe design and implementation of the Gridware System framework, then optimizes andimproves the fusion of the Gridware System, increases a variety of field fusions according tothe actual needs. On the other hand, this paper improves the engine of the Gridware System,in other words, the execution environment of the Gridware System, including the parsingengine and execution engine, increases operations such as analysising and performing of thedistributed code. The Gridware System becomes a distributed computing environment, andcan be used for the development and implementation of a distributed Gridware program.