HBase storage schemas for massive spatial vector data

Abstract

With the development of Geographic Information System (GIS), the storage requirement of spatial vector data is increasing dramatically. Nowadays, designing an efficient storage schema for massive spatial vector data becomes a key step for GIS. Cloud computing with NoSQL, such as HBase, can provide massive high-concurrent and scalable service for storage of spatial vector data. However, storage schemas in NoSQL for spatial vector data can be rarely seen. In this paper, two HBase storage schemas for spatial vector data are proposed. One is the storage schema with rowkeys based on Z curve, Z schema, and the other is the storage schema with rowkeys based on geometry objects identifiers, ID schema. In our experiments, the region query efficiency of the two storage schemas is tested on the cloud framework built by us. Different order Z curve and different query ranges are involved in the experiments. Experimental results show, for both schemas, the increase of query range leads to the growth of response time. More importantly, response time of Z schema is about one-fifth as long as that of ID schema in all cases. It can be seen that Z schema is a better solution for storing spatial vector data in HBase.