Query Performance Comparison of PostgreSQL vs. Neo4j. A Basic Distributed Setup on OpenStack

Among the NoSQL technologies, Neo4j is one of the most popular solutions for managing graph databases and an early adopter of transactions (contrary to other NoSQL Systems). Neo4j also provides a powerful high-level data processing language - Cypher. Despite its popularity, there are few comprehensive studies about benchmarking the query performance of Neo4j relative to SQL or other NoSQL counterparts. In this paper, we present a module for converting the TPC-H benchmark database from PostgreSQL to Neo4j, and we built a set of 110 SQL queries that were translated into Cypher. For both database servers, the queries were executed with a 10-minute timeout on OpenStack setups following nine scenarios by combining three database scale factors $(1 \mathrm{~GB}, 5 \mathrm{~GB}$, and 10 GB) with three data distribution variants (with 1,5, and 10 nodes). Results provide support for query performance assessment of these two big data products.