En prestandajämförelse av Neo4j och SQL Server utifrån spatiala användningsområden

Abstract

Värderingsdata is a company which specializes in evaluating different types of properties in Sweden. To accomplish high accuracy in their evaluations, a considerable amount of data as well as complex and time consuming queries are utilized. In an effort to remedy the time aspect, this report’s aim has been to determine the advantages and disadvantages of the currently-in-use relational database SQL Server and the graph database Neo4j. Furthermore, their individual performance is to be determined. The different advantages and disadvantages of the databases will be assessed in conjunction with tests as will their respective performance. The tests in question consist of several different queries based on three different scenarios. These are centered around retrieving a propertiey’s closest point and area of interest as well as any relevant spatial information. The results from the tests conducted showed the relational database as performing better in all scenarios. Neo4j’s lack of native support for spatial functionality was one of its downsides. This in combination with its performance compared to the relational database lead to the conclusion that Neo4j in its current state is not a viable substitution to SQL Server.