Research Of The Architecture Of Hadoop And Its Application In Medical Insurance Balance System

Along with the development of computing and networking technology and their application in medical insurance, to carry out medical insurance business is increasingly dependent on the support of the medical insurance system. The trend of expansion of medical insurance business overall planning scope is also resonated with the new direction of current cloud computing technology. Therefore the study of architecture of cloud computing platform, the investigation of software development in cloud environment and the combination of cloud computing research and medical insurance application are of great practical significance.Hadoop is a recently emerged excellent open-source cloud computing platform and its good scalability and outstanding performance in massive data processing attracted many researchers’ attention. The goal of this paper is to use Hadoop open-source framework for building experiment cloud computing platform cluster enviroment, then develop and deploy medical insurance application software on built Hadoop cluster and compare the performance of system deployed in MapReduce parallel computing approach with the performance of system developed in usual B/S model. All we do is to provide reference for developers of software.Fisrstly, this paper studies the architecture of Hadoop thoroughly, focuses on HDFS and MapReduce including their architecture and running mechanism and so on, researches the MapReduce programming model and the steps of executing a MapReduce program and compares Hadoop platform with other well-known cloud computing platforms.Based on the penetrating understanding of the architecture of Hadoop, this paper introduces how to build Hadoop clusters and how to develop MapReduce programs. Then this paper analyzes the information needs of medical insurance business, designs and implements medical insurance balance system in two different approaches that one is based on the traditional B/S mode and the other based on Hadoop MapReduce pattern. Finally, this paper compares the performance differences between the two approaches. The experiment results show that powerful computing and storage capacity of Hadoop cluster emerges when it completes large data processing tasks and verify the design concept of Hadoop.