In order to comply with the rapid development of our data resources, this paper focuses on the collection, analysis and integration of our multi-source heterogeneous data resources based on the data-lake structure (short for the forward data flow), and publishes the standard data from our data-lake to many characteristic field centers through the micro service (short for the reverse data flow). So as to keep the data fresh by bidirectional data flow, and to achieve the optimal utilization rate. It effectively supports the data display and knowledge service of many characteristic field centers, and supports scientific researchers with convenience of utilizing knowledge resources in their professional field. At present, the bidirectional data flow mechanism has established friendly cooperation with more than 10 characteristic field centers, and has provided regular and continuous data flow services, the amount of data reached to 18 million.