Created: November 2019
Description
Neo4j can be substantially faster for certain types of queries because of the benefits gained from index-free adjacency.
Example: Get the names of destination airports from all flights originating in Wyoming
Cypher (Neo4j query language):
MATCH (hi:Airport {state: 'WY'})-[:HAS_DEPARTURE]->(fl:Flight)-[:FLIES_TO]->(ap:Airport)
Postgres SQL:
RETURN DISTINCT ap.name
SELECT name from airports
JOIN flights ON (airports.iata = flights.destination_airport)
WHERE flights.origin_airport IN
(SELECT iata from airports WHERE airports.state = 'WY')
| Attempt | 1 | 2 | 3 | Average | Logo |
|---|---|---|---|---|---|
| Neo4j | 0.29222047s | 0.34756797s | 0.30003158s | 0.32918392s |  |
| PostgreSQL | 2.32982001s | 2.14832110s | 2.39080437s | 2.29873976s |  |
Context
This performance comparison was completed for the final project of my SCS 3252:017 Big Data Management Systems & Tools course (University of Toronto).
There is a PDF and Jupyter notebook investigating how Neo4j can outperform PostgreSQL for large interconnected datasets. See below:
Technology Stack
- neo4j
- PostgreSQL
- jupyter notebooks
- Google Colab
Links
- Project PDF Document: Comparing Neo4j with PostgreSQL ⎘
- Jupyter Notebook: Flightly Comparing Neo4j and PostgreSQL.ipynb ⎘
- Github: Github ⎘
