Traditional relational database management systems (RDBMS) have shown limitations in storing and analyzing big data. For example, a RDBMS is suitable for transactional operations yet not good at large-scale data analysis and processing, since a large-scale record scan or full table scan is often time-consuming. An efficient storage method for reading and writing big geospatial data is still needed. In this paper, we propose a geospatial data storage method based on the HBase and MapReduce. There are two key parts in our storage method. First, we present a MapReduce based refinement method for geospatial data access to improve I/O efficiency. Second, we design a structure of HBase tables to efficiently store and manage geospatial data. As a result, compared to traditional RDBMS, the storage model provides a better solution for high-perfermance writing and storage of geospatial data.
展开▼
