Grid computing is a powerful resources sharing architecture with great potential. However, with the great number of resources, assigning jobs to grid sites with suitable resources is very difficult. Each job needs a different combination of resources. In this paper, we propose a decision tree approach for the system manager to monitor how many combinations of resources are actually used. Our method allows the system manager to decide if there are too many, which combinations should stay and which should be upgraded or reallocated? The method is shown to be simple and effective.