PROVIDER Β· AWS fully managed graph database service
Amazon Neptune
Amazon Neptune is AWS's fully managed graph database supporting property graphs via Gremlin and openCypher plus RDF via SPARQL, with three-AZ storage replication, 15 read replicas, and Neptune Analytics for in-memory analysis.
4.1pts
Pros
Cons
Product Information
Detailed Notes
Overview
Amazon Neptune became generally available in 2018 as AWS's fully managed graph database supporting property graphs through Gremlin and openCypher plus RDF through SPARQL, with storage replicated across three availability zones, up to 15 read replicas, and Neptune Analytics for in-memory graph analysis.
For Backend & Database work the value is operations-free graph capability, since no cluster setup or tuning is needed and instance hours buy a highly available graph store integrated with IAM, VPC, CloudWatch, and SageMaker, while two graph models let RDF knowledge graphs and property graphs share one stack.
Watch cost and performance traits, because instance hours plus memory sensitivity mean small sizes slow down clearly, write throughput trails in-memory engines, deep traversal performance depends on query shape, and cross-cloud migration rewrites queries and models.
Core Strengths
- Supports property graphs via Gremlin and openCypher plus RDF via SPARQL, covering both ecosystems
- Storage replicates across three AZs with up to 15 read replicas and automatic failover
- Deep integration with AWS IAM, VPC, CloudWatch, and SageMaker
- Neptune Analytics adds in-memory analysis and built-in algorithms without a self-built compute cluster
Product Ecosystem
Neptune Database
The main service with Gremlin, openCypher, and SPARQL billed per instance size and storage.
Neptune Analytics
An in-memory analysis engine loading from S3 or Neptune and running built-in algorithms.
Neptune Serverless
Capacity-unit autoscaling suited to development and unpredictable load.
Drawbacks
- Instance-hour billing costs more long term than self-hosted open source graphs
- Write throughput and very deep traversal trail in-memory dedicated engines
- Runs only inside AWS, so cross-cloud migration rewrites queries and models
- Algorithm capability lives in Neptune Analytics while the main database is limited
Use Cases
- Knowledge graphs on AWS (β β β β β ) - Native SPARQL and RDF
- Relationship graphs and recommendation paths (β β β β ) - Gremlin and openCypher
- Operations-free graph analytics (β β β β ) - Fully managed with Analytics
- Extreme write throughput and deep links (β β β ) - In-memory engines fit better
Pricing
| Plan | Price | Scope |
|---|---|---|
| Neptune Serverless | Per capacity unit | Autoscaling for intermittent load |
| Neptune instances | Per instance-hour | Storage and I/O billed separately |
| Neptune Analytics | Per memory and m-NCU hour | In-memory analysis loaded on demand |
Neptune bills instance hours plus storage and I/O, and since queries are memory sensitive an undersized instance slows down and raises total cost, so batch algorithm work suits Neptune Analytics loaded on demand rather than a large always-on instance.
FAQ
-
Neptune or Neo4j? Neptune for full AWS adoption needing operations-free hosting with SPARQL and RDF, Neo4j for the mature Cypher ecosystem, stronger write performance, and visualisation.
-
Which query languages are supported? Gremlin and openCypher for property graphs, SPARQL for RDF, with one graph model per cluster and no mixing.
-
Can it run graph algorithms? The database provides querying only, so algorithms run in Neptune Analytics with built-in PageRank, community detection, and similarity, or after export elsewhere.