NoSQL spatial – Neo4j versus PostGIS

The relational data model is now more than 40 years old. It's good for very different scenarios and can handle certain types of data very well. But it isn't perfect. An increasing number of non-relational systems are being worked on, collectively called the 'NoSQL movement'. NoSQL is an umbrella term for a loosely defined class of non-relational datastores and best described as 'Not only SQL'. Examples are Google's BigTable and Amazon's Dynamo. These systems may provide advantages over relational databases, but generally lack the relational robustness for those advantages. In an attempt to modestly contribute to the current research in the field of geographical information systems, this research reports on a comparison of one such NoSQL datastore called Neo4j with the traditional relational database PostgreSQL when storing and querying spatial vector data. The goal of this study was to determine whether a traditional relational database system like PostGIS, or a NoSQL system, such as Neo4j-Spatial, would be more effective as the underlying technology for operating OpenStreetMap data. Neo4j is an embeddable graph database. Embeddable because it can be added to a Java-based application and used just like any other library, and graph database because the data model it uses to express its data is a graph, storing nodes and relationships that connect them, supporting user defined properties on both constructs. Multiple aspects are of interest when comparing spatial database systems. The evaluation methodology designed to compare the two, involves both objective measurements and subjective measurements based on documentation and experience. The objective tests include processing speed based on a predefined set of queries, disk space requirements, and scalability. Subjective tests include maturity/level of support, stability and ease of use. Meanwhile, the assessment framework applied for the objective tests could be used as a test suite for evaluating the performance and