Graph Databases Landscape and Comparison Cheat Sheet

Updated 2026-05-16

Next Topic: InfluxDB and Time Series Databases Cheat Sheet

🧠Study flashcards on this topic151 cards · spaced repetition→

Graph databases excel at modeling and querying highly connected data through nodes, edges, and relationships, treating connections as first-class citizens rather than expensive joins. Unlike relational databases that require multiple table scans for traversals, native graph storage enables constant-time relationship lookups, making them uniquely suited for social networks, fraud detection, knowledge graphs, and recommendation engines. The field has evolved into two primary paradigms: property graphs (supporting Cypher/Gremlin queries over labeled nodes and edges with attributes) and RDF graphs (using SPARQL to query semantic web triples), with modern systems increasingly bridging both through multi-model support and the emerging GQL ISO standard.

What This Cheat Sheet Covers

This topic spans 22 focused tables and 153 indexed concepts, 151 flashcards. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.

Table 1: Leading Graph Database PlatformsTable 2: Graph Data ModelsTable 3: Query Languages for GraphsTable 4: Gremlin Traversal Steps (Apache TinkerPop)Table 5: SPARQL Query Forms (W3C Standard for RDF)Table 6: Property Graph vs RDF Data ModelTable 7: Graph Database Use CasesTable 8: Graph Schema Design PatternsTable 9: Graph Algorithms for AnalyticsTable 10: Graph Embeddings for Machine LearningTable 11: Query Optimization and PerformanceTable 12: Backup, Recovery, and High AvailabilityTable 13: Security and Access ControlTable 14: Data Import and ETL StrategiesTable 15: Graph Database Architecture PatternsTable 16: CAP Theorem and Consistency ModelsTable 17: Temporal Graphs and VersioningTable 18: Open Source Licensing and EditionsTable 19: Identity Resolution and Entity LinkingTable 20: Cypher Query Language Syntax (Neo4j)Table 21: Vector Search and Graph Hybrid (RAG Integration)Table 22: Graph Database Monitoring and Observability

Quick IndexSubscribe to unlock

A jump-to index of every table row in this cheat sheet.

Mind MapSubscribe to unlock

An interactive map of every table and concept in this topic.

Table 1: Leading Graph Database Platforms

The graph database market splits along two lines that matter when you pick one: whether the engine is purpose-built for graphs or layered on existing storage, and which query language it speaks. Neo4j and TigerGraph optimize for deep native traversals, ArangoDB and NebulaGraph emphasize multi-model or distributed scale, and Amazon Neptune is the managed cloud option that hedges across both property-graph and RDF worlds.

Database	Example	Description
Neo4j	Native property graph with `MATCH (p:Person)-[:FRIENDS_WITH]->(f) RETURN f.name`	• Most widely adopted graph database supporting Cypher query language • offers both GPLv3 Community Edition and commercial Enterprise Edition with clustering, backup, and advanced security features
Amazon Neptune	Managed graph on AWS supporting `g.V().has('name','Alice').out('knows')` (Gremlin)	• Fully managed service supporting both property graphs (Gremlin, openCypher) and RDF graphs (SPARQL 1.1) • provides automatic backups, point-in-time recovery, and multi-AZ replication on AWS infrastructure
ArangoDB	Multi-model database with `FOR v IN 1..3 OUTBOUND 'users/alice' GRAPH 'social'` (AQL)	• Combines document, key-value, and graph models in a single engine • uses AQL (ArangoDB Query Language) for unified querying across models with native JSON document storage alongside graph traversals
TigerGraph	Native parallel graph with `SELECT src, tgt FROM Person:s-(KNOWS:e)->Person:t` (GSQL)	• Designed for real-time deep-link analytics with massively parallel processing • uses GSQL query language and excels at multi-hop traversals across billions of edges with sub-second latency
JanusGraph	Distributed graph using `g.V().has('user','name','alice').out('created')`	• Open-source (Apache 2.0) distributed graph built on pluggable storage backends (Cassandra, HBase, Berkeley DB) and indexing systems (Elasticsearch, Solr) • supports TinkerPop Gremlin traversals
NebulaGraph	Distributed graph with separation of storage/compute using `GO FROM "player101" OVER serve` (nGQL)	• Open-source distributed architecture with graph, storage, and meta services separated • designed for horizontal scalability with billions of vertices and uses nGQL (NebulaGraph Query Language).
Dgraph	Native GraphQL database using `{ queryPerson(filter: {name: {eq: "Alice"}}) { friends { name } } }`	• Horizontally scalable with native GraphQL support and DQL (Dgraph Query Language) • provides ACID transactions and automatic sharding with focus on developer-friendly GraphQL-first API

Table 2: Graph Data Models

Before choosing a database you have to choose how connections are actually represented. The property-graph family attaches key-value attributes directly to nodes and edges and reads naturally for application developers, while RDF breaks everything into subject-predicate-object triples built for the semantic web. The remaining entries are variations and blends — multiple labels, hyperedges that join more than two things, and multi-model engines that store documents alongside the graph.

Model	Example	Description
Property graph (LPG)	Nodes `(:Person {name:"Alice", age:30})` with edges `[:KNOWS {since:2020}]`	• Nodes and edges both carry key-value properties • nodes tagged with labels, edges have types • optimized for application-centric modeling with fast traversals and flexible schema evolution

The rest of this section is part of the full cheat sheet

Table 3: Query Languages for Graphs

There is no single SQL for graphs yet, which is why the same traversal looks wildly different across products. Cypher and its open-source twin openCypher lead the property-graph side and feed the new GQL ISO standard, Gremlin offers a vendor-neutral step-by-step traversal style, and SPARQL remains the W3C standard for RDF. The vendor-specific languages below (AQL, nGQL, DQL, GSQL) trade portability for features tuned to their own engine.

Language	Example	Description
Cypher	`MATCH (a:Person)-[:KNOWS]->(b) WHERE a.age > 25 RETURN b.name`	• Declarative pattern-matching language for property graphs • ASCII-art syntax makes relationship patterns visually intuitive • developed by Neo4j, now part of openCypher standard

The rest of this section is part of the full cheat sheet

Table 4: Gremlin Traversal Steps (Apache TinkerPop)

Gremlin builds a query by chaining small steps, so it helps to know the vocabulary you reach for most. These cover the full lifecycle of a traversal — entering the graph with V()/E(), hopping across edges with out/in/both, filtering and projecting, then looping, pathing, and aggregating to turn raw walks into answers.

Step	Example	Description
addV / addE	`g.addV('person').property('name','alice')` `g.V(1).addE('knows').to(V(2))`	• Creates new vertices or edges in the graph • addV creates nodes with specified labels and properties • addE adds relationships between vertices with direction

The rest of this section is part of the full cheat sheet

Table 1: Leading Graph Database Platforms

Database	Example	Description
Neo4j	Native property graph with `MATCH (p:Person)-[:FRIENDS_WITH]->(f) RETURN f.name`	• Most widely adopted graph database supporting Cypher query language • offers both GPLv3 Community Edition and commercial Enterprise Edition with clustering, backup, and advanced security features
Amazon Neptune	Managed graph on AWS supporting `g.V().has('name','Alice').out('knows')` (Gremlin)	• Fully managed service supporting both property graphs (Gremlin, openCypher) and RDF graphs (SPARQL 1.1) • provides automatic backups, point-in-time recovery, and multi-AZ replication on AWS infrastructure
ArangoDB	Multi-model database with `FOR v IN 1..3 OUTBOUND 'users/alice' GRAPH 'social'` (AQL)	• Combines document, key-value, and graph models in a single engine • uses AQL (ArangoDB Query Language) for unified querying across models with native JSON document storage alongside graph traversals
TigerGraph	Native parallel graph with `SELECT src, tgt FROM Person:s-(KNOWS:e)->Person:t` (GSQL)	• Designed for real-time deep-link analytics with massively parallel processing • uses GSQL query language and excels at multi-hop traversals across billions of edges with sub-second latency
JanusGraph	Distributed graph using `g.V().has('user','name','alice').out('created')`	• Open-source (Apache 2.0) distributed graph built on pluggable storage backends (Cassandra, HBase, Berkeley DB) and indexing systems (Elasticsearch, Solr) • supports TinkerPop Gremlin traversals
NebulaGraph	Distributed graph with separation of storage/compute using `GO FROM "player101" OVER serve` (nGQL)	• Open-source distributed architecture with graph, storage, and meta services separated • designed for horizontal scalability with billions of vertices and uses nGQL (NebulaGraph Query Language).
Dgraph	Native GraphQL database using `{ queryPerson(filter: {name: {eq: "Alice"}}) { friends { name } } }`	• Horizontally scalable with native GraphQL support and DQL (Dgraph Query Language) • provides ACID transactions and automatic sharding with focus on developer-friendly GraphQL-first API

Table 2: Graph Data Models

Model	Example	Description
Property graph (LPG)	Nodes `(:Person {name:"Alice", age:30})` with edges `[:KNOWS {since:2020}]`	• Nodes and edges both carry key-value properties • nodes tagged with labels, edges have types • optimized for application-centric modeling with fast traversals and flexible schema evolution

Table 3: Query Languages for Graphs

Language	Example	Description
Cypher	`MATCH (a:Person)-[:KNOWS]->(b) WHERE a.age > 25 RETURN b.name`	• Declarative pattern-matching language for property graphs • ASCII-art syntax makes relationship patterns visually intuitive • developed by Neo4j, now part of openCypher standard

Table 4: Gremlin Traversal Steps (Apache TinkerPop)

Step	Example	Description
addV / addE	`g.addV('person').property('name','alice')` `g.V(1).addE('knows').to(V(2))`	• Creates new vertices or edges in the graph • addV creates nodes with specified labels and properties • addE adds relationships between vertices with direction