Sequential pattern mining analyzes the ordered user behaviors, such as ordered list of the products purchased by most of the users. Because the user transactions will be increased every day, the current sequential patterns may be different from the previous ones. Therefore, how to efficiently update the original sequential patterns in real time is a very important research topic. If the original sequential patterns cannot be updated in time, then the information may no longer represent the user behaviors. For the previous studies in this area, some approaches may loss information, and some methods need to refind the previous discovered patterns. In this article, we propose a novel approach for mining and maintaining the discovered sequential patterns without losing any information and re-discovering the existed patterns. The experiments also represent that our approach is more efficient than the current most efficient algorithm.