Skip to content

Commit 691d35f

Browse files
POTUSAITEJAPOTUSAITEJA
committed
Add 7 more databases and improve discoverability with SEO metadata
1 parent 7d6a060 commit 691d35f

12 files changed

Lines changed: 565 additions & 20 deletions

File tree

‎README.md‎

Lines changed: 24 additions & 16 deletions
Original file line numberDiff line numberDiff line change
@@ -2,19 +2,21 @@
22

33
# Graph Database Resource Catalog
44

5-
### A curated, primary-source reference for 47 graph databases across RDF, property graph, multi-model, and specialized systems — with explicit vector-search classification.
5+
### The curated, primary-source reference for graph databases, RDF triple stores, property graph systems, and vector-capable graph engines.
6+
7+
**A comprehensive comparison of 54 graph databases — Neo4j, Apache Jena, Stardog, ArangoDB, Memgraph, TigerGraph, Dgraph, Virtuoso, GraphDB, Amazon Neptune, Oracle Graph, Google Spanner Graph, and more — with explicit vector-search classification for every entry.**
68

79
<p>
810
<a href="#license"><img alt="License" src="https://img.shields.io/badge/License-MIT-34d399?style=for-the-badge"></a>
911
</p>
1012

1113
<p>
12-
<img alt="Entries" src="https://img.shields.io/badge/Entries-47-60a5fa?style=flat-square">
13-
<img alt="RDF" src="https://img.shields.io/badge/RDF-11-60a5fa?style=flat-square">
14-
<img alt="LPG" src="https://img.shields.io/badge/LPG-18-fb923c?style=flat-square">
15-
<img alt="Multi-model" src="https://img.shields.io/badge/Multi--model-13-22d3ee?style=flat-square">
16-
<img alt="Specialized" src="https://img.shields.io/badge/Specialized-5-c084fc?style=flat-square">
17-
<img alt="Native vector" src="https://img.shields.io/badge/Native%20vector-19-34d399?style=flat-square">
14+
<img alt="Entries" src="https://img.shields.io/badge/Entries-54-60a5fa?style=flat-square">
15+
<img alt="RDF" src="https://img.shields.io/badge/RDF-13-60a5fa?style=flat-square">
16+
<img alt="LPG" src="https://img.shields.io/badge/LPG-19-fb923c?style=flat-square">
17+
<img alt="Multi-model" src="https://img.shields.io/badge/Multi--model-15-22d3ee?style=flat-square">
18+
<img alt="Specialized" src="https://img.shields.io/badge/Specialized-7-c084fc?style=flat-square">
19+
<img alt="Native vector" src="https://img.shields.io/badge/Native%20vector-21-34d399?style=flat-square">
1820
</p>
1921

2022
<p>
@@ -31,11 +33,17 @@
3133

3234
---
3335

34-
## What this is
36+
## About
37+
38+
The **Graph Database Resource Catalog** is a curated, primary-source reference for every graph database, RDF triple store, SPARQL engine, label property graph database, multi-model engine, and graph-plus-vector system worth considering for production use. It answers one question honestly:
39+
40+
> **Which graph databases actually support vector search, and how?**
41+
42+
Every one of the 54 entries is anchored to official documentation, product pages, or source repositories. The catalog distinguishes native vector support from connector-based, paired-extension, and sibling-service patterns — and where the public docs are ambiguous, the entry is marked `unclear` instead of being inflated.
3543

36-
A production-oriented reference that answers one question honestly: **which graph databases actually support vector search, and how?**
44+
**Topics covered:** graph database · RDF database · SPARQL · property graph · label property graph · LPG · knowledge graph · semantic graph · triple store · Cypher · Gremlin · GQL · SQL/PGQ · Datalog · vector search · HNSW · GraphRAG · multi-model database · graph analytics · graph OLTP · graph OLAP · knowledge graph engineering · semantic web
3745

38-
Every entry is anchored to primary sources (official product pages, documentation, or source repositories). The catalog distinguishes native vector support from connector-based, sibling-service, and paired-extension patterns. Where public docs are ambiguous, the entry is marked `unclear` instead of being inflated.
46+
**Databases in the catalog include:** Neo4j, Apache Jena, Fuseki, Virtuoso, Stardog, GraphDB, Amazon Neptune, RDFox, MarkLogic, Oxigraph, QLever, Blazegraph, TerminusDB, SurrealDB, ArangoDB, ArcadeDB, Memgraph, TigerGraph, Dgraph, JanusGraph, NebulaGraph, Apache HugeGraph, Apache AGE, OrientDB, TypeDB, Azure Cosmos DB for Gremlin, PuppyGraph, Kuzu, Ultipa, FalkorDB, Aerospike Graph, Tarantool Graph DB, TuGraph, AgensGraph, Fluree, RedisGraph, CozoDB, Oracle Graph, Google Spanner Graph, Gel (formerly EdgeDB), Datomic, HyperGraphDB, AnzoGraph DB, Weaviate, Cayley, DataStax Enterprise Graph, Eclipse RDF4J, AtomicServer, HelixDB, Dydra, and more.
3947

4048
## What makes it different
4149

@@ -63,21 +71,21 @@ Every entry is anchored to primary sources (official product pages, documentatio
6371

6472
**By model**
6573

66-
- **11** RDF / SPARQL-native stores
67-
- **18** Label property graph databases
68-
- **13** Multi-model systems with a real graph layer
69-
- **5** Specialized graph-adjacent systems
74+
- **13** RDF / SPARQL-native stores
75+
- **19** Label property graph databases
76+
- **15** Multi-model systems with a real graph layer
77+
- **7** Specialized graph-adjacent systems
7078

7179
</td>
7280
<td valign="top" width="50%">
7381

7482
**By vector strategy**
7583

76-
- **19** Native (in-engine)
84+
- **21** Native (in-engine)
7785
- **4** Integrated
7886
- **5** Paired
7987
- **2** Sibling-service
80-
- **14** None
88+
- **19** None
8189
- **3** Unclear
8290

8391
</td>

‎databases/atomic-server/README.md‎

Lines changed: 43 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,43 @@
1+
# AtomicServer
2+
3+
> Reference implementation of the Atomic Data specification: a type-safe, RDF-inspired graph database with JSON ergonomics, built in Rust.
4+
5+
| | |
6+
|---|---|
7+
| **Website** | [atomicserver.eu](https://atomicserver.eu) |
8+
| **Model** | Specialized |
9+
| **Status** | Emerging |
10+
| **License** | MIT (permissive) |
11+
| **Query** | REST / Atomic queries |
12+
| **Vector strategy** | **None** |
13+
| **HA** | Single-node (sled-backed storage) |
14+
| **Deployment** | Self-hosted, single binary |
15+
16+
## Overview
17+
18+
AtomicServer is the reference implementation of the Atomic Data specification, a data model that combines RDF-style typed links, JSON compatibility, and runtime type safety. Every resource has a URL, a declared schema, and typed properties, which makes the data layer feel like a lightweight linked-data graph while still being easy to work with from JavaScript and TypeScript clients.
19+
20+
The server is written in Rust, backed by the sled key-value store, and ships as a single small binary. It also doubles as a headless CMS with a built-in table editor, full-text search, and client SDKs for JavaScript, React, Svelte, and Rust.
21+
22+
## Vector strategy — None
23+
24+
No native vector index is documented. AtomicServer focuses on typed, linked data rather than similarity search. Applications that need vector search alongside an Atomic Data store would pair it with a separate vector engine at the application layer.
25+
26+
## Best fit
27+
28+
- Headless CMS and real-time collaborative data applications
29+
- Projects that want a type-safe, RDF-inspired graph without the ceremony of a full SPARQL stack
30+
- Small-to-medium deployments where a single-binary store is operationally attractive
31+
- Front-end teams that prefer JSON-compatible resources over classical RDF serialisations
32+
33+
## Considerations
34+
35+
- Single-node storage; not aimed at distributed deployments
36+
- No SPARQL surface — queries go through the REST API and Atomic queries
37+
- Emerging project status; ecosystem and tooling are still growing
38+
- No vector search
39+
40+
## Sources
41+
42+
- [Atomic Data documentation](https://docs.atomicdata.dev/)
43+
- [AtomicServer GitHub repository](https://github.com/atomicdata-dev/atomic-server)

‎databases/cayley/README.md‎

Lines changed: 43 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,43 @@
1+
# Cayley
2+
3+
> Open-source linked-data graph database inspired by the Google Knowledge Graph, written in Go with pluggable backend stores.
4+
5+
| | |
6+
|---|---|
7+
| **Website** | [cayley.io](https://cayley.io) |
8+
| **Model** | Multi-model |
9+
| **Status** | Production |
10+
| **License** | Apache 2.0 (permissive) |
11+
| **Query** | Gizmo (JavaScript DSL) / GraphQL / MQL |
12+
| **Vector strategy** | **None** |
13+
| **HA** | Depends on the chosen backend store |
14+
| **Deployment** | Self-hosted with LevelDB, BoltDB, SQL, PostgreSQL, or MongoDB backends |
15+
16+
## Overview
17+
18+
Cayley is an open-source graph database inspired by the graph engine behind Google's Knowledge Graph, formerly Freebase. It is written in Go and ships with a pluggable backend so the same engine can sit on top of LevelDB, BoltDB, PostgreSQL, SQL databases, or MongoDB depending on the deployment profile.
19+
20+
The project started inside Google, was open sourced in 2014, and is now maintained by the community under the cayleygraph organisation on GitHub. It is typically used for linked-data and RDF-shaped workloads where a lightweight Go-native engine is preferred over a heavier Java stack.
21+
22+
## Vector strategy — None
23+
24+
No native vector index is documented. Cayley predates the modern vector-database era and focuses on linked-data and triple-shaped storage. Applications that need vector search alongside Cayley's graph model would need to pair it with a separate vector engine at the application layer.
25+
26+
## Best fit
27+
28+
- Open-source linked-data projects that want a Go-native engine
29+
- Teams who need to choose a storage backend (embedded key-value vs. existing SQL or NoSQL)
30+
- Research or prototyping with multiple query languages (Gizmo, GraphQL, MQL)
31+
- Small-to-medium linked-data deployments that do not require a full enterprise triple store
32+
33+
## Considerations
34+
35+
- Ecosystem tooling is smaller than mature RDF and LPG stores
36+
- No native vector capability
37+
- HA and durability characteristics depend on the backend chosen
38+
- Active but community-maintained; release cadence is slower than commercial products
39+
40+
## Sources
41+
42+
- [Cayley GitHub repository](https://github.com/cayleygraph/cayley)
43+
- [Cayley project site](https://cayley.io)

‎databases/databases.json‎

Lines changed: 147 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1016,6 +1016,153 @@
10161016
"https://www.cambridgesemantics.com/anzograph/"
10171017
],
10181018
"notes": "AnzoGraph DB is the MPP SPARQL engine embedded inside the broader Anzo platform from Cambridge Semantics; also marketed as Altair Graph Studio after the Altair acquisition."
1019+
},
1020+
{
1021+
"id": "weaviate",
1022+
"name": "Weaviate",
1023+
"model": "specialized",
1024+
"status": "production",
1025+
"license": "BSD-3-Clause",
1026+
"license_type": "permissive",
1027+
"website": "https://weaviate.io",
1028+
"query_interfaces": ["GraphQL", "REST", "gRPC"],
1029+
"vector_strategy": "native",
1030+
"vector_summary": "Weaviate is a vector-first database with HNSW indexes and typed cross-references between objects that form a navigable knowledge graph.",
1031+
"ha": "Sharding and replication across cluster nodes",
1032+
"deployment": "Self-hosted, Docker, Kubernetes, or Weaviate Cloud",
1033+
"best_fit": "Vector-first knowledge graphs with GraphQL access and schema-defined cross-references",
1034+
"profile": "databases/weaviate/README.md",
1035+
"source_urls": [
1036+
"https://github.com/weaviate/weaviate",
1037+
"https://docs.weaviate.io/weaviate"
1038+
],
1039+
"notes": "Primarily a vector database; the graph dimension comes from typed cross-references between objects rather than arbitrary edges. Catalogued as specialized because it is not a classical LPG or RDF engine."
1040+
},
1041+
{
1042+
"id": "cayley",
1043+
"name": "Cayley",
1044+
"model": "multi",
1045+
"status": "production",
1046+
"license": "Apache 2.0",
1047+
"license_type": "permissive",
1048+
"website": "https://cayley.io",
1049+
"query_interfaces": ["Gizmo (JavaScript DSL)", "GraphQL", "MQL"],
1050+
"vector_strategy": "none",
1051+
"vector_summary": "No native vector index; Cayley is a linked-data graph database inspired by the Google Knowledge Graph and Freebase with pluggable backend stores.",
1052+
"ha": "Depends on the chosen backend store",
1053+
"deployment": "Self-hosted with LevelDB, BoltDB, SQL, PostgreSQL, or MongoDB backends",
1054+
"best_fit": "Open-source linked-data projects that want a Go-native engine with multiple query languages and pluggable storage",
1055+
"profile": "databases/cayley/README.md",
1056+
"source_urls": [
1057+
"https://github.com/cayleygraph/cayley",
1058+
"https://cayley.io"
1059+
],
1060+
"notes": "Google-origin project inspired by the Freebase / Knowledge Graph architecture. Maintained by the community under cayleygraph/cayley on GitHub."
1061+
},
1062+
{
1063+
"id": "dse-graph",
1064+
"name": "DataStax Enterprise Graph",
1065+
"model": "property",
1066+
"status": "production",
1067+
"license": "Commercial",
1068+
"license_type": "commercial",
1069+
"website": "https://www.datastax.com/products/datastax-graph",
1070+
"query_interfaces": ["Gremlin", "CQL"],
1071+
"vector_strategy": "none",
1072+
"vector_summary": "No vector capability is documented in the DSE Graph engine; vector features live in DataStax's separate Astra DB product line.",
1073+
"ha": "Cassandra-native multi-datacenter replication",
1074+
"deployment": "Self-hosted DataStax Enterprise cluster",
1075+
"best_fit": "Enterprise graph workloads backed by Apache Cassandra's distributed storage and linear scale-out",
1076+
"profile": "databases/dse-graph/README.md",
1077+
"source_urls": [
1078+
"https://docs.datastax.com/en/dse/6.9/graph/graph-contents.html",
1079+
"https://www.datastax.com/products/datastax-graph"
1080+
],
1081+
"notes": "Built on top of Apache Cassandra inside DataStax Enterprise. Classic DSE Graph and the newer DataStax Graph engine share the same product surface from DSE 6.8 onward."
1082+
},
1083+
{
1084+
"id": "rdf4j",
1085+
"name": "Eclipse RDF4J",
1086+
"model": "rdf",
1087+
"status": "production",
1088+
"license": "EDL 1.0 / EPL 2.0",
1089+
"license_type": "permissive",
1090+
"website": "https://rdf4j.org",
1091+
"query_interfaces": ["SPARQL 1.1", "SeRQL"],
1092+
"vector_strategy": "none",
1093+
"vector_summary": "No native vector index; RDF4J is a Java framework whose bundled Native Store and Memory Store provide embeddable SPARQL 1.1 endpoints.",
1094+
"ha": "Single-node stores; clustering depends on the SAIL backend",
1095+
"deployment": "Embedded Java library, standalone RDF4J Server, or Workbench",
1096+
"best_fit": "Java applications that need an embeddable SPARQL 1.1 store or a framework for building one",
1097+
"profile": "databases/rdf4j/README.md",
1098+
"source_urls": [
1099+
"https://rdf4j.org/about/",
1100+
"https://projects.eclipse.org/projects/technology.rdf4j"
1101+
],
1102+
"notes": "Formerly known as OpenRDF Sesame. Ships both an on-disk (Native Store) and in-memory (Memory Store) SAIL implementation, plus a pluggable SAIL architecture for third-party backends."
1103+
},
1104+
{
1105+
"id": "atomic-server",
1106+
"name": "AtomicServer",
1107+
"model": "specialized",
1108+
"status": "emerging",
1109+
"license": "MIT",
1110+
"license_type": "permissive",
1111+
"website": "https://atomicserver.eu",
1112+
"query_interfaces": ["REST", "Atomic queries"],
1113+
"vector_strategy": "none",
1114+
"vector_summary": "No native vector index; AtomicServer is the reference implementation of the Atomic Data specification, a type-safe, RDF-inspired, JSON-compatible graph data model.",
1115+
"ha": "Single-node (sled-backed storage)",
1116+
"deployment": "Self-hosted, single binary",
1117+
"best_fit": "Headless CMS and real-time collaborative data apps that want a type-safe, RDF-inspired graph model",
1118+
"profile": "databases/atomic-server/README.md",
1119+
"source_urls": [
1120+
"https://docs.atomicdata.dev/",
1121+
"https://github.com/atomicdata-dev/atomic-server"
1122+
],
1123+
"notes": "Reference implementation of the Atomic Data specification. The underlying data model combines RDF-style links with JSON ergonomics and runtime type safety."
1124+
},
1125+
{
1126+
"id": "helixdb",
1127+
"name": "HelixDB",
1128+
"model": "multi",
1129+
"status": "emerging",
1130+
"license": "AGPL-3.0",
1131+
"license_type": "copyleft",
1132+
"website": "https://www.helix-db.com",
1133+
"query_interfaces": ["HelixQL"],
1134+
"vector_strategy": "native",
1135+
"vector_summary": "HelixDB is a Rust-native graph-vector database that ships both graph traversal and vector search as first-class primitives inside HelixQL, its strongly-typed compiled query language.",
1136+
"ha": "Single-node (LMDB-backed storage)",
1137+
"deployment": "Self-hosted; managed service available for selected users",
1138+
"best_fit": "RAG and AI applications that need graph traversal and vector search in a single typed query surface",
1139+
"profile": "databases/helixdb/README.md",
1140+
"source_urls": [
1141+
"https://github.com/HelixDB/helix-db",
1142+
"https://www.helix-db.com"
1143+
],
1144+
"notes": "Rust implementation with LMDB storage. AGPL-3.0 has implications for SaaS distribution; review the license before embedding in commercial services."
1145+
},
1146+
{
1147+
"id": "dydra",
1148+
"name": "Dydra",
1149+
"model": "rdf",
1150+
"status": "production",
1151+
"license": "Commercial",
1152+
"license_type": "commercial",
1153+
"website": "https://dydra.com",
1154+
"query_interfaces": ["SPARQL 1.1", "GraphQL", "Linked Data Platform"],
1155+
"vector_strategy": "none",
1156+
"vector_summary": "No native vector index is documented; Dydra is a versioned cloud RDF store with first-class temporal snapshots and SPARQL 1.1 support.",
1157+
"ha": "Managed cloud (recent work integrates the RonDB distributed backend)",
1158+
"deployment": "Managed cloud service operated by Datagraph",
1159+
"best_fit": "Cloud-native RDF applications that need versioned, time-travelable SPARQL stores",
1160+
"profile": "databases/dydra/README.md",
1161+
"source_urls": [
1162+
"https://docs.dydra.com/dydra",
1163+
"https://docs.dydra.com/sparql"
1164+
],
1165+
"notes": "A 'Git for graphs' approach — every store state is fully addressable via a REVISION argument in SPARQL. Recent work with Hopsworks/RonDB targets trillion-triple scale."
10191166
}
10201167
]
10211168
}

‎databases/dse-graph/README.md‎

Lines changed: 42 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,42 @@
1+
# DataStax Enterprise Graph
2+
3+
> Enterprise property graph engine built on top of Apache Cassandra as part of DataStax Enterprise, queried with Gremlin.
4+
5+
| | |
6+
|---|---|
7+
| **Website** | [datastax.com/products/datastax-graph](https://www.datastax.com/products/datastax-graph) |
8+
| **Model** | Label Property Graph |
9+
| **Status** | Production |
10+
| **License** | Commercial |
11+
| **Query** | Gremlin / CQL |
12+
| **Vector strategy** | **None** |
13+
| **HA** | Cassandra-native multi-datacenter replication |
14+
| **Deployment** | Self-hosted DataStax Enterprise cluster |
15+
16+
## Overview
17+
18+
DataStax Enterprise Graph, usually abbreviated as DSE Graph, is the property graph layer inside DataStax Enterprise. It stores graph data in Apache Cassandra tables and exposes Gremlin as the primary query language, letting teams use Cassandra's distributed storage, linear scale-out, and multi-datacenter replication for graph workloads that need continuous availability.
19+
20+
From DSE 6.8 onward the classic DSE Graph engine and the newer DataStax Graph engine share the same product surface. Both target the same use case: running production graph workloads on top of Cassandra without operating a separate graph database.
21+
22+
## Vector strategy — None
23+
24+
No vector capability is documented inside the DSE Graph engine itself. DataStax ships vector search as a distinct product line (Astra DB and related Cassandra vector features) rather than as a graph engine feature, so applications that need vector search alongside DSE Graph typically pair it with a separate DataStax product or an external vector engine.
25+
26+
## Best fit
27+
28+
- Enterprise graph workloads that already depend on Apache Cassandra for distributed storage
29+
- Multi-datacenter deployments that require continuous availability and linear scale-out
30+
- Teams that want Gremlin over a Cassandra-backed storage layer with a commercial support contract
31+
32+
## Considerations
33+
34+
- Commercial-only licensing and enterprise procurement cycle
35+
- No native vector support in the graph engine
36+
- Gremlin-only query surface; no native Cypher or SPARQL
37+
- Tied to the DataStax Enterprise release train and support model
38+
39+
## Sources
40+
41+
- [DataStax Graph documentation](https://docs.datastax.com/en/dse/6.9/graph/graph-contents.html)
42+
- [DataStax Enterprise Graph product page](https://www.datastax.com/products/datastax-graph)

0 commit comments

Comments
 (0)