GDB-Engines

The go-to resource for graph technology details

About
Back to all databases

Neptune

A fully managed graph database service from AWS supporting both property graph and RDF models. Available as Neptune Database (for transactional workloads) and Neptune Analytics (for analytical queries).

Visit website

Elsewhere

Overall rank
#14
First released
November 2017
License
Proprietary
Built with
C++

Details

Vendor
Amazon Web Services
Type
Multiple
Category
Enterprise
License
Proprietary
Language
C++
Query Languages
GremlinSPARQLopenCypher
Protocols
BoltHTTP/RESTSPARQL ProtocolGremlin (WebSocket)Gremlin (HTTP)

Lineage

Bigdata2007BlazeGraphendedNeptune2017
  • BlazeGraph: renamed from Bigdata; no longer developed.
  • Neptune, 2017: based on BlazeGraph.

Feature scores

Community & Business

Active development
✓
Updated within last 3 months
Commercial support
✓
Paid support available
Live community
✓
>75% of issues addressed
Open Source
0%
Closed source
Pricing
✓
Pricing publicly available
Trendiness
✓
Growing or stable trend

Deployment

Containerization
0%
No container image
Work as dedicated instance
✓
Can run as standalone server
Work as embedded
0%
Standalone only
Testing in-memory version
0%
No in-memory option

Platform

Operating on Linux
✓
Linux officially supported
Operating on Windows
0%
Not supported
SaaS offering
✓
Cloud service available

Operations

Automatic updates
✓
Automatic updates supported
Client side caching
✓
Client-side cache available
Data versioning support
0%
No versioning
Live backups
✓
Hot backup supported

Distribution

Cluster Re-balancing
0%
No auto-rebalancing
Data Distribution
0%
No sharding
High-Availability
✓
HA mechanisms available
Query Distribution
50%
No description available
Replication support
✓
Replication available

Developer Experience

Data types defined
50%
Basic data types only
Logging/Auditing
✓
Event logging supported
Object-Graph Mapper
0%
No object mapper
Reactive programming
✓
Reactive streams supported
Documentation up-to-date
✓
Docs current with latest version
Binary protocol
✓
Binary protocol available
CLI
✓
Command-line interface
GUI
✓
Visual interface available

Data Model

Multi-database
0%
Graph-only
Graph-native data
✓
Native graph storage
REST API
✓
REST endpoint available
Query Language
✓
Standard query language (Cypher/Gremlin)

Transactions

Granular locking
✓
Object-level locking
Multiple isolation levels
✓
Multiple isolation levels
Read committed transaction
✓
Read committed available
Transaction support
✓
ACID transactions

Schema & Security

Constraints
✓
Full constraint support
Schema support
50%
Constraints/triggers only
Secondary indexes
✓
Secondary indexes supported
Server side procedures
0%
No stored procedures
Triggers
0%
No triggers
Authentication
✓
Authentication supported
Authorization
✓
Role-based access control
Data encryption
✓
Data encryption supported
Sources 12 references

Protocols Bolt, Gremlin (WebSocket), Gremlin (HTTP), SPARQL Protocol, HTTP/REST

Bolt is TCP-only on Neptune (the WebSocket variant of the spec is not supported); Gremlin has both a WebSocket driver channel and a separately documented HTTPS REST endpoint; SPARQL is served over its own 1.1 Protocol HTTP endpoint. Kept HTTP/REST because it is independently justified by the openCypher HTTPS endpoint (a generic REST channel with no more specific vocabulary tag) and by the separate Loader/management HTTP API, not just as a stand-in for SPARQL or Gremlin. Rejected pass_B's drop of Gremlin (HTTP) and HTTP/REST.

License Proprietary

Neptune is a closed-source, fully managed AWS commercial service with no published source code or license file, so it is classified as Proprietary rather than under any open-source SPDX id.

Released 2017-11

Matches current value; a limited preview was launched at re:Invent on November 29, 2017, well ahead of the May 2018 general availability.

Language C++

AWS does not directly disclose Neptune's implementation language. Amazon's own paper on Poseidon, the engine behind Neptune Analytics, describes a C++ template-heavy codebase, and Neptune engineering job postings require C++ (or Rust) proficiency for the storage/compute core, so C++ is the best-supported inference for the primary engine language.

  • For these reasons, our code relies heavily on C++ templates. The templates implement the access patterns as defined and enumerated by the P8APL interface.
    Poseidon: A OneGraph Engine · reviewed 2026-08-02
  • 1+ years of experience in C++, Rust, Related programming language- Knowledge of database internals, query planning and optimization and/or storage performance and efficiency
    Software Development Engineer, Amazon Neptune - Job ID · reviewed 2026-08-02

Lineage based on BlazeGraph

AWS has never said so itself. The link rests on the Blazegraph team (including its CEO), trademark and domain moving to AWS in the year before Neptune launched.