A novel method to manage very large raster data on distributed key-value storage system
Yunqin Zhong, Shangchun Sun, Haojun Liao, Yanwei Zhao, Jinyun Fang · 2011
With the rapid development of information technologies and GIS techniques, raster data amount is growing on an unprecedented scale. Existing WebGIS based on local file systems and RDBMS cannot manage very large raster data efficiently because of limited storage capacity of single node. Although expensive storage devices are used to enlarge capacity, WebGIS is still vulnerable to suffer from single point of failure for its poor scalability. We propose a novel method to manage very large raster data, it has three characteristics. Firstly, raster data management system is built upon distributed key-value storage system instead of local file system and RDBMS, it has good scalability and high availability. Secondly, revised quadtree-based raster index is built in system to improve access efficiency. Thirdly, our prototype is transparent to WebGIS applications, and hence can integrate with existing applications seamlessly. The experimental results show that our method outperforms existing methods.