Recently, all kinds of cloud computing applications are popular since the network bandwidth raises. The company needs to set up the server to maintain the data in the earlier year. Nowadays, the company just rents the virtual server from the cloud service provider, and then it can flexibly adjust the server level according to its requirement. Moreover, this scheme can save the deployment and maintenance costs of the company. However, it calls for an efficient mechanism to shorten the task completion time when the amount of data is waiting to process in data center networks. Unfortunately, it is an NP-complete problem in the Hadoop system for minimum task completion time. Recently, many studies propose improvement schemes, but these schemes still can be enhanced. Therefore, in this paper, we proposed a bandwidth-aware rescheduling (BARE) mechanism in the software-defined network (SDN)-based data center networks. BARE uses the feature of the SDN to collect the global information, sorts out the locality data process, performs the task split, and executes the rescheduling mechanism for reducing the task completion time. Finally, we use the simulation to prove our scheme. The performance results show that our scheme outperforms other existing schemes.