Skip to content

Specialised databases

Beyond the relational, key-value and warehouse services, AWS sells a set of databases each built for one shape of data. They are worth knowing about mostly so that you recognise the workload when you meet it: a graph query written against a relational schema, or time-series data stored a row at a time, is usually slow for reasons no amount of tuning will fix.

This family is also the part of the AWS database catalogue that has moved most in recent years, so check a service’s current status before designing around it.

  • Amazon Neptune — graph database, for data whose relationships are themselves the thing being queried: identity graphs, recommendations, fraud detection, network and security topology.
  • Amazon DocumentDB — managed document database with MongoDB compatibility, for JSON documents with flexible schemas.
  • Amazon Keyspaces — managed, serverless Apache Cassandra-compatible wide-column store.
  • Amazon Timestream — time-series data. Read this one before adopting it: the engine most material describes is closed to new customers, and Timestream for InfluxDB is what AWS now sells.
  • Amazon QLDB — the managed ledger database. Support ended on 31 July 2025; the page explains what it did and what replaces it.

Two neighbours are frequently filed under “specialised databases” and are not databases at all. Amazon Athena is a SQL query engine over S3 and is covered under big data processing and analytics. Amazon OpenSearch Service — renamed from Amazon Elasticsearch Service in September 2021 — is a search and log analytics engine and is covered under observability.