Based on the architecture of the cloud computing, we establish a distributive CNC system, and analysis differences of the task request between the general cloud computing and the CNC system, then we propose a dynamic feedback scheduling-algorithm in order to realize tasks scheduling. In this dynamic feedback algorithm, every task has a weight value, which is expressed by a vector, in order to express the different request of the different resource. Quantization analysis of the resource request is used to select the best node to accomplish the task, and the optimization selection can achieves the purpose that improving the usage of the resource and reducing the average response time of tasks. Lastly, the simulation results verify the feasibility and the availability of the algorithm in distributive algorithm.