HydraDB vs Amazon Neptune
Neptune for RDF or an AWS-native service, but its engine is closed and runs only on AWS. HydraDB is open source, can run in your AWS account, and builds the graph from Slack, email and docs, not just S3 files.
Based on public docs and pricing as of Nov 2026. Something out of date? Suggest a correction →
Neptune is a managed service that handles both property graphs and RDF, with Gremlin, openCypher and SPARQL. It's the go-to for standards-based knowledge graphs, and for teams whose graph has to live inside AWS with IAM, VPC and Bedrock and no new vendor to approve.
The friction shows up when you need to know what's happening under the hood, or when the data your agents need isn't sitting in S3.
The caveats worth knowing
You can't look under the hood. The engine is closed. When a query is slow or an answer looks wrong, you can't read the code to find out why. It runs only on AWS, with no self-hosted version or official local build, so teams develop against emulators like LocalStack. Leaving means a full export.
Managed doesn't mean built for you. AWS runs the servers. You still pick a data model and query language, then shape your data into CSV or RDF files and bulk-load them from S3.
Automatic graph building only reads S3. Bedrock can build a Neptune graph from documents, but only from S3, with no control over how it's built. Slack, email and tickets have to be exported there first.
What changes with HydraDB?
The engine is open source. You can read the code, debug against it and run it on a laptop. It stores the graph on S3 and can run in your own AWS account, so choosing it doesn't mean leaving AWS.
The graph builds from your sources. Entities and relationships are created as data arrives, structured to your ontology or one proposed for you, and the graph updates incrementally as new facts come in. There are no files to prepare or bulk loads to schedule.
Connect 3 apps, export 1 table, keep 1 export fresh, then ask.
Design the schema first, export 3 sources, map 3 sources onto your schema, run 3 loads, set up extraction for 1 source, then write queries.
Need RDF, or a graph that has to live inside AWS with IAM, VPC and Bedrock? Neptune is the pick.
Sources connect directly. 100+ connectors keep Slack, Gmail, Drive, Jira and others in sync, permissions included, with no S3 staging step.
Technical comparison
| Amazon Neptune | HydraDB | |
|---|---|---|
| Architecture | Purpose-built managed engine; storage grows automatically up to 128 TiB | Written in Rust; the graph lives on object storage, with memory and NVMe as caches |
| Scaling | One primary instance handles all writes, with up to 15 read replicas; Serverless scales within limits you set | Compute and storage scale separately |
| Data models | Property graph and RDF | Property graph |
| Building the graph | Bulk load CSV or RDF from S3; Bedrock GraphRAG from S3 documents | 100+ app connectors and file uploads, structured to your ontology or a proposed one, then updated incrementally |
| When a fact changes | New value replaces the old | Every version kept, Git-style |
| Deployment | AWS only: provisioned, Serverless or Analytics | Managed, dedicated, your cloud, or self-hosted |
| Pricing | Instances from $0.093/hr (db.t4g.medium, 4 GiB) to $72.30/hr (db.x2iedn.32xlarge, 4 TiB), or Serverless at $0.16 per capacity-unit hour; plus $0.10 per GB-month storage and $0.20 per million I/O requests on Standard (pricing) | Managed from $25/month: $0.70 per 1M tokens ingested, $0.50 per GB-month storage; open source free (pricing) |
| License | Proprietary | Elastic License 2.0 |