Aiming at the characteristics of data center traffic and the problem of a single traffic classification reference metric, this paper proposes a data center traffic classification method based on the triangular mode fusion. The article describes the characteristics of data center traffic, and then it redefines the membership functions of the flow size and the flow cut-off time, and then the membership functions are merged based on the triangular mode fusion algorithm, so as to determine the priority of traffic more reasonably and improve the performance efficiency and user experience of data center.