Abstract: In view of the low exchange of big data shared resource in libraries domestic and foreign, and the lack of one-stop service platform crossing all standalone system, this paper his paper is focused on improving the sharing and utilization of digital resources from the technical aspect, proposes a structure model for the sharing of library digital resource based on Hadoop to solve the problem of low sharing performance and the lack of systematic management. Moreover, it provides readers with a multi-faceted service.The system uses MapReduce distributed computing framework and a resource management system to process digital resource data in parallel to achieve fast correlation search and processing of digital resources, and introduces a composite library big data storage system based on HDFS which can guarantee the storage requirements of the resource sharing system, effectively improve the access efficiency for the readers, and solve the scalability problems faced by the traditional storage technologies.

Key words: digital resource sharing, Hadoop, MapReduce, HDFS