Manufacturing service process is a manufacturing service chain based on business process, which have four basic structures:sequence, selection, loop and parallelism. Since the loop can be transformed into sequential structure, the selection structure and parallel structure are real branch structures. Branch structures often have service capability difference, with the results that the selection branch will delay time due to improper probability allocation, while the parallel branch will be in a waiting state, which will inevitably affect the overall execution efficiency of business process. A time-improved manufacturing service flow optimization method with QoS assurance was proposed in this paper. Firstly, the attribute calculation method of the basic structure of business process was constructed, and then by analyzing the influence relationship between the time of several branch structures and other attribute factors, the layering and blocking linear programming model of branch structure time optimization based on QoS constraint was established, and the hierarchical and block optimization algorithm was designed. An example of the process to be optimized was selected in the experiment, and the execution time of the optimized business process was reduced by 5.4%, which showed that the model and its optimization algorithm are effective and reasonable. This study has positive significance for the application of cloud manufacturing.