Abstract:

A descriptive methodology and its aided construction tool based on feature model for collaborative computing tasks over multiple virtual machines (CCTMVM) are studied to simple the description of parallel computing in virtualized platform and improve the development efficiency of applications. The tasks decomposition policies are discussed from the principle and method of tasks decomposition, tasks particle judgment and the heuristic rules, and then the appraisal criterion of parallel relationships among tasks is determined to describe the semantic actions of feature units based on their timing relationships. The validity of the descriptive method is determined to judge the effectiveness of the methodology presented from the coefficients of cohesion coupling and the granularities of feature units. A prototype of aided construction tools for CCTMVM is designed, and then a problem-solving application in computer aided engineering (CAE) is taken for an example to analyze its task decomposition. The experimental results show that the proposed idea and methodology have a certain feasibility for constructing a parallel computingoriented CCTMVM.