Software defined network (SDN) is an emerging network paradigm that proposes to decouple the control and data forwarding plane. Leveraging the centralized control ability and open programming interfaces SDN provides, network management can be dramatically simplified, and network services can be dynamically controlled by applications. Traffic engineering (TE) is a category of mechanisms that promises to optimize network performance through analyzing, predicting and regulating data flow in network. The unique features of SDN provide TE with unified and real-time global network view, flexible control manner and better policy for scalable traffic management, exhibiting profound research significance. This paper surveys the state-of-art works on TE for SDN. In terms of the methods, application and deployment of measurement, works on traffic measurement architecture, network invariant detection and measurement resource management with SDN are reviewed. The problems in traditional network traffic management are analyzed, and SDN-based data traffic management methods for efficient load balance and control traffic management methods for centralized bottleneck relief are presented. Moreover, network failure resilience technologies in SDN are described. Finally, the technological approaches are summarized and future research issues are discussed.