Neo4j
| Company type | Private |
|---|---|
| Industry | Software, Database management |
| Founded | 2007 |
| Headquarters | San Mateo, California; Malmö , United States; Sweden |
| Products | Neo4j graph database system |
| Website | neo4j |
Neo4j is a graph database system engineered by Neo4j Inc.
This database manages information through nodes, the edges that link them, and their corresponding properties. Neo4j operates as an ACID-compliant transactional database, featuring native graph storage alongside processing capabilities. It provides a non-open-source "community edition" governed by a modified GNU General Public License, while advanced features such as online backup and high availability are distributed under proprietary commercial licenses. These commercial extensions are also licensed by Neo4j Inc.
Constructed using Java, Neo4j enables interaction from multiple programming languages via the Cypher query language, accessible through either a transactional HTTP interface or the binary "Bolt" protocol. The suffix "4j" alludes to its Java roots, though this designation is now largely considered outdated.
History
Neo4j is created and maintained by Neo4j, Inc., which has offices located in San Mateo, California, United States, and Malmö, Sweden.
The product's first official release, Neo4j 1.0, occurred in February 2010. Notable subsequent major versions include 2.0, released in December 2013, and 3.0, which debuted in April 2016.
The company’s growth has been marked by key funding rounds: a $36 million Series D led by Greenbridge Partners Ltd in November 2016; an $80 million Series E round spearheaded by One Peak Partners and Morgan Stanley Expansion Capital in November 2018; followed by a substantial $325 million Series F investment announced in June 2021.
Release history
| Version | Initial release[1] | Most recent minor update[2] | Latest version[2] | Support termination date[1] |
|---|---|---|---|---|
| 1.0 | 2010-02-23 | Template:Version | Template:NA | 2011-08-23 |
| 1.1 | 2010-07-30 | Template:Version | Template:NA | 2012-01-30 |
| 1.2 | 2010-12-29 | Template:Version | Template:NA | 2012-06-29 |
| 1.3 | 2011-04-12 | Template:Version | Template:NA | 2012-09-12 |
| 1.4 | 2011-07-08 | Template:Version | Template:NA | 2013-01-08 |
| 1.5 | 2011-11-09 | Template:Version | Template:NA | 2013-03-09 |
| 1.6 | 2012-01-22 | Template:Version | Template:NA | 2013-07-22 |
| 1.7 | 2012-04-18 | Template:Version | Template:NA | 2013-10-18 |
| 1.8 | 2012-09-28 | Template:Version | Template:NA | 2014-03-28 |
| 1.9 | 2013-05-21 | Template:Version | 2014-10-13 | 2014-11-21 |
| 2.0 | 2013-12-11 | Template:Version | 2014-07-08 | 2015-06-11 |
| 2.1 | 2014-05-29 | Template:Version | 2015-04-01 | 2015-11-29 |
| 2.2 | 2015-03-25 | Template:Version | 2016-06-16 | 2016-09-25 |
| 2.3 | 2015-10-21 | Template:Version | 2017-12-12 | 2017-04-21 |
| 3.0 | 2016-04-16 | Template:Version | 2017-10-03 | 2017-10-31 |
Licensing and editions
Neo4j is available in five distinct versions: two designed for on-premises deployment—Community (which is free) and Enterprise—and three exclusively cloud-based options: AuraDB Free, AuraDB Professional, and AuraDB Enterprise.
The platform employs a dual-licensing model, combining GPL v3 (with certain components under AGPLv3 alongside Commons Clause) and a proprietary license. The Community Edition is offered at no cost but is limited to a single-node setup without support for clustering or hot backup capabilities.
In contrast, the Enterprise Edition lifts these limitations by providing features such as clustering, hot backups, and monitoring tools, distributed under a closed-source commercial license.
Neo4j’s licensing strategy has sparked controversy and legal disputes.
Data structure
Data in Neo4j is structured using nodes, the relationships (edges) that link these nodes, and their associated properties. Both nodes and relationships can be tagged with labels to enhance query precision. Starting with version 2.0, Cypher introduced schema-based indexing, which boosts query efficiency compared to the earlier approach of separate indexes.
Criticisms
Database specialist Andy Pavlo has expressed skepticism regarding the transition from traditional relational databases to specialized graph models. Additionally, researchers at CWI conducted benchmarks comparing a modified version of DuckDB against Neo4j for graph-related workloads, demonstrating that their SQL-based relational extension outperforms Neo4j in specific scenarios.
Neo4j initiated legal action against PureThink for removing the Commons Clause from its AGPL license. In 2024, a U.S. District Court ruled in favor of Neo4j, ordering PureThink to pay $597,000 in damages for violations of licensing terms and trademark infringement concerning Neo4j branding in sales to government entities.