Recently

cloud technology is widely used in Data Center Network（DCN）.In order to ensure Quality of Service（QoS）in DCN

Equal Cost Multipath Routing（ECMP）algorithms and Dynamic Load Balancing（DLB）algorithms are employed as solutions.However

these algorithms only obtain local optimization results.In particular

the transmission delay and bandwidth utilization are not satisfactory when these algorithms are applied in the DCN overloaded environment.Considering the limitations of these algorithms

this paper proposes a load balancing scheduling mechanism for fat tree network data traffic under Software Defined Network（SDN）architecture from the perspective of global optimization.The mechanism evaluates the SDN node and link load by monitoring the global real-time parameters through the Ryu controller

and selecting the optimal forwarding path which meets the demand of traffic load.Experiments show that the scheduling mechanism proposed in this paper has significantly improved the performance of DCN services compared with the ECMP/DLB algorithm.