Autonomic database system is a performance optimization system, which can improve the resource utilization of database system by effective management of the access order to the workload of the database system. In this paper, the existing workload autonomic framework of database system has been improved, giving emphasis on the workload control component within the framework, and the comparative analysis of the replacement strategy and the NSGA-II algorithm was carried out. The simulation shows that the NSGA-II workload control algorithm can achieve higher accuracy with scarifying longer time to run, whereas the replacement strategy of workloads has lower accuracy with less time-consuming, which is more favorable for a real-time autonomic database system.