Software-defined wide area network (SD-WAN) is a new type of network architecture that has developed rapidly in recent years. SD-WAN inherits the centralized control ar-chitecture of Software-defined networking (SDN), but supports more diverse access methods and equipment types and covers a wider area. It is also associated with greater uncertainty in the network environment. These characteristics make the fault management of the SD-WAN controller more challenging, so that the existing SDN-based data backup methods cannot adapt to SD-WAN scenarios. This paper proposes an SD-WAN-oriented Distributed Data Backup and Recovery method (DDBR) based on an improved secret sharing algorithm. To deploy this method, we design an online-offline dual backup framework based on the data freshness requirements of the controller. Under this framework, dynamic data of the controller is divided into different shares, and then stored into the storage of switches. When the controller fails, data recovery can be performed on the backup controller quickly, which greatly improves the network availability. The outstanding feature of the proposed DDBR method is that it ensures the integrity and confidentiality of the backup data in an unreliable network environment, even when some of the storage nodes fail. Evaluation results on file backup example show that the proposed solution has significant advantages over existing methods in terms of backup data storage size and backup success rate.