HDFS enabled storage and management of remote sensing data

Abstract

The continuously growing volume of massive remote sensing data raised huge challenges on storage space and querying efficiency. In this paper, a new management model of remote sensing data has been proposed to address these issues of system architecture, data storage strategies, and functionalities based on Hadoop Distributed File System (HDFS). The model is capable of relieving the overloading problems of the single NameNode server in HDFS by taking a dual storage mechanism and a simulating operations method. On one hand, Relational Database Management System (RDBMS) and HDFS are separately taken to store and manage image files and related metadata of remote sensing data; on the other hand, operations of file systems are simulated by RDBMS. The study results show the model could improve management and storage efficiency of remote sensing data.