Lesson 1: Every AWS Database Service on One Page

Why start with a map

AWS sells more than a dozen managed database services. Several of them have been renamed, and a few are closed to new customers. Teams lose real time by choosing a service that has just been retired, or by searching for a name the console no longer uses. So before we compare anything, we build a one-page map. It lists every service, the group it belongs to, the names it has had and whether you can still sign up for it.

Think of it as a hardware store floor plan. You don't need to know every tool yet, but you should know which aisle sells screws and which shelf is being cleared out.

The seven aisles

Gotcha: "closed to new customers" is not the same as "shut down". Accounts that already use Timestream for LiveAnalytics can keep running it. A new account, or a new team that has never used it, can't start. For new time-series work, AWS points you to Timestream for InfluxDB.

The renames that confuse people

ElastiCache. Redis changed its licence in 2024, and the Linux Foundation community forked Redis OSS 7.2.4 as Valkey under the BSD 3-clause licence. AWS now names the service plain Amazon ElastiCache and treats Valkey, Redis OSS and Memcached as engines you pick. Some existing Redis OSS clusters can be upgraded to Valkey in place. MemoryDB made the same change: it now supports Valkey as well as Redis OSS, and Valkey launched at a lower price.

Timestream. "Timestream" now covers two different products. LiveAnalytics is the original serverless engine, which is closed to new customers. InfluxDB is managed open-source InfluxDB running on instances you size. If a blog post from 2022 says "Timestream", it means LiveAnalytics.

Schema conversion. SCT is the desktop tool you install. DMS Schema Conversion is the managed, web-based version inside DMS. It supports fewer source and target platforms, so you still need SCT for data warehouse schemas, application SQL and ETL code.

Walkthrough: see what your account already runs

The map tells you what exists. These read-only CLI calls tell you what your account uses. Each command belongs to a different service namespace, and that list is itself a quick way to remember the map.

REGION=eu-west-1   # pick your Region; most of these services are Regional

# Relational
aws rds describe-db-instances --region $REGION \
  --query 'DBInstances[].[DBInstanceIdentifier,Engine]' --output table
# Aurora clusters are listed under the rds namespace too
aws rds describe-db-clusters --region $REGION \
  --query 'DBClusters[].[DBClusterIdentifier,Engine]' --output table

# Distributed SQL
aws dsql list-clusters --region $REGION

# Key-value and document
aws dynamodb list-tables --region $REGION
aws docdb describe-db-clusters --region $REGION --query 'DBClusters[].DBClusterIdentifier'

# Wide-column
aws keyspaces list-keyspaces --region $REGION

# In-memory
aws elasticache describe-replication-groups --region $REGION \
  --query 'ReplicationGroups[].[ReplicationGroupId,Engine]'
aws memorydb describe-clusters --region $REGION --query 'Clusters[].[Name,Engine]'

# Graph and time series
aws neptune describe-db-clusters --region $REGION --query 'DBClusters[].DBClusterIdentifier'
aws timestream-influxdb list-db-instances --region $REGION

# Oracle Database@AWS (namespace: odb)
aws odb list-cloud-exadata-infrastructures --region $REGION

# Migration
aws dms describe-replication-instances --region $REGION

Here is typical output for an account that has some history:

-------------------------------------
|        DescribeDBInstances        |
+-------------------+---------------+
|  orders-prod      |  postgres     |
|  legacy-erp       |  custom-oracle-ee |   <-- RDS Custom for Oracle: plan a move before 31 Mar 2027
+-------------------+---------------+

Best practice: Run this inventory in every Region you use and keep the results with your selection records. An Engine value starting with custom-oracle should go on your risk list now, not next spring.

Pitfall: The docdb and neptune commands call the same underlying API as RDS. If you filter by engine name you'll get the right results. If you run aws rds describe-db-clusters with no filter, the list can include DocumentDB and Neptune clusters, which surprises people.

Edge cases on the map

• Limitless versus DSQL. Both scale SQL out across many nodes, so we file both under distributed SQL. Limitless is a feature of an Aurora PostgreSQL cluster. DSQL is a separate, serverless, multi-Region service.
• Oracle Database@AWS is relational, but it is not RDS. Oracle runs the Exadata service on hardware in AWS data centres. That changes who you call for support and how billing works, and later lessons cover what that means for you.
• Ledger workloads have no dedicated service any more. With QLDB gone, an audit-trail design now means an append-only table in Aurora PostgreSQL or another service, not a ledger product.

Key takeaways

• Seven groups: relational, distributed SQL, key-value and document, wide-column, in-memory, graph and time series, and migration.
• ElastiCache and MemoryDB are now service names with engines to choose: Valkey, Redis OSS, or Memcached (ElastiCache only).
• Timestream is two products. LiveAnalytics closed to new customers on 20 June 2025, and InfluxDB is the path AWS recommends.
• QLDB reached end of support on 31 July 2025. RDS Custom for Oracle is in sunset and ends on 31 March 2027.
• Keyspaces was previewed as Amazon Managed Apache Cassandra Service (MCS).
• DMS moves the data. DMS Schema Conversion (managed) or SCT (desktop, broader coverage) converts the schema.
• Check the service's status page before you design around it, and record the date you checked.