Although the commercial software like CAD/CAM can be adopted to handle most machining problems, the tedious manual procedures are still indispensable. We developed a rule-based feature recognition technique which can be used to recognize common features of the drilling holes. Furthermore, we apply the TSP (Traveling Sales Person) model to simulate the optimal drilling tool-path. Result shows TSP is able to improve both of the manual configuration time and drilling time. In order to verify this method, an industry-specific injection mold was selected as an example. The proposed method can successfully identify all the holes and compare the results of several different optimal paths to obtain the most suitable method for the mold.