Abstract: The control system of high intensity heavy ion accelerator facility (HIAF) adopts EPICS (experimental physics and industrial control system) architecture. The running data archive system is supposed to provide a reliable storage system for the accelerator’s running data, historical data and static parameters. The archive system based on EPICS architecture was redesigned as the new HIAF Archive Engine. A MongoDB interface to support NoSQL database was developed using MongoDB’s embedded data model and multilevel document storage to optimize table structures. And a more effective method to store waveform and image data was provided. Then a manual archive solution for accelerator operators based on Flag PV was added, thus to improve the query efficiency and increase the historical data’s usability. To meet the requirements for massive historical data storage, a distributed database cluster based on MongoDB sharding solution was provided.