To solve the problems of frequent collision of elephant flows, severe link congestion and unbalanced load in data center networks, a multi-path traffic scheduling algorithm based on Ant Colony Optimization (ACO) is proposed.The algorithm combines the multi-path characteristics of network topology with the advantages of Software Defined Network (SDN) global view.First, the large and small flows are detected by using the sFlow controller.Then, the K short path algorithm is used to add the non-shortest paths on the basis of the shortest path, and calculate the criticality of each path in the path library.The less critical path is selected as the candidate path to reduce the cost of the scheduling algorithm.Finally, according to the characteristics of the large and small flows, the link transmission cost and the pheromone's update method in the ACO are set respectively, and the multi-path traffic scheduling is realized by calculating the number of optimal paths and assigning weights.Simulation results show that under different-type traffic modes, the proposed algorithm effectively improves the average network bisection bandwidth, lowers the maximum link utilization, and reduces the average transmission delay.