Research And Implementation Of Massive Medical Data Storage System Based On Hadoop

With the fast and sound development of information technology in healthcare, medical data rapidly emerge in large numbers. However, existing platforms for patients’data storage cannot meet the needs of the even increasing large volume of medical data. Therefore, it is very important to develop effective storage platform to manage and store these massive medical data.Cloud computing offers low cost, high scalability, availability and fault tolerance which provides a good solution for some of the problems faced in storing and analyzing patients’ medical data. Based on the distributed computing technology, this paper proposes a novel approach for mass medical data storage and management. It includes a solution for storing massive medical data storage platform based on Hadoop by using Linux cluster technology.Based on the research of the cloud storage and Hadoop technologies at home and abroad, this paper designs a new medical data storage system based on Hadoop. This system consists of three parts:storage center, management center and application center. Data storage center uses HBase as its database, and is supported by the underlying distributed file system (HDFS).This paper improves the original load balancing algorithm of HDFS and applies the new multi-indexed algorithm to the management center of the medical storage system. This algorithm is intended for controlling the load distribution and migration processes of the cluster.Based on the design of the system, this paper also elaborates on the implementation of massive medical data storage system based on Hadoop. We set up a Hadoop cluster environment and design a management software in the application center. The data storage, visualization, retrieval and other functions are realized in this software. At last, the stress testing experiments are carried out, and the results show that the system are well-performed in load balancing control under the circumstance of heavy system load.In the end, the work of this paper is summarized and the future work is discussed.