AWS Database Migration Service (DMS)
AWS Database Migration Service (DMS) is a cloud service of Amazon Web Services used to securely migrate databases from on-premises, Amazon EC2, or Amazon RDS to AWS. It does not stop the running application while performing the migration, minimizing downtime, and it is the standard answer for moving an on-premises MySQL or PostgreSQL database to Amazon RDS.
Key points
- Homogeneous migration — same database engine on both sides, e.g. Oracle DB (on-premises) → AWS Database Migration Service → Amazon RDS for Oracle.
- Heterogeneous migration — different database engines on each side, converting the schema before migrating the data, e.g. an on-premises Oracle schema converted for Amazon Aurora, then migrated with AWS DMS → Amazon Aurora.
- Schema conversion now lives in two places. The original standalone, downloadable AWS Schema Conversion Tool (AWS SCT) desktop application still exists and still supports the broadest range of source/target engine combinations. AWS has since added DMS Schema Conversion directly inside the DMS console — a browser-based experience (no separate install) covering the most common source/target pairs (e.g. Oracle, SQL Server, PostgreSQL, MySQL, Db2 LUW, SAP ASE as sources; Amazon Aurora, Amazon RDS, and Amazon Redshift as targets). Use the console-native DMS Schema Conversion for common paths; fall back to standalone AWS SCT when it supports an engine pair DMS Schema Conversion doesn’t.
- Supported source engines (data migration): Oracle, Microsoft SQL Server, MySQL, MariaDB, PostgreSQL, MongoDB, SAP ASE, IBM Db2 LUW, IBM Db2 for z/OS, Amazon Aurora, Amazon S3, Amazon DocumentDB, and Amazon RDS for Db2 LUW, plus Azure SQL/PostgreSQL/MySQL and Google Cloud MySQL/PostgreSQL as third-party managed sources.
- Supported target engines (data migration): Oracle, Microsoft SQL Server, MySQL, MariaDB, PostgreSQL, SAP ASE, IBM Db2 LUW, Amazon Aurora (including PostgreSQL Limitless and Serverless v2), Amazon Redshift (including Redshift Serverless), Amazon S3, Amazon DynamoDB, Amazon OpenSearch Service, Amazon ElastiCache (Redis OSS), Amazon Kinesis Data Streams, Amazon DocumentDB, Amazon Neptune, Apache Kafka (including Amazon Managed Streaming for Apache Kafka (MSK)), and Babelfish for Aurora PostgreSQL.
- DMS Serverless removes replication-instance management entirely: AWS automatically provisions, scales, and patches the compute capacity (measured in DMS capacity units) for a migration or ongoing replication task based on CloudWatch CPU/capacity utilization, and (as of an April 2025 update) also auto-scales storage — so there’s no more capacity planning, instance sizing, or fixed 100 GB storage ceiling to manage.
- Performs all management steps required during migration, including monitoring, scaling, error handling, network connectivity, replicating during failure, and software patching.
- Scope is wider than relational: it also migrates data warehouses, NoSQL databases and other data stores, including an on-premises NoSQL workload into Amazon DynamoDB.
- Migrations can run cloud-to-cloud, on-premises-to-cloud, or between combinations of the two — though one endpoint must always be an AWS service; DMS can’t migrate directly between two on-premises databases.
- Distinct from the server-level migration services — AWS Server Migration Service (now retired) and its successor AWS Transform MGN move whole virtual machines, while DMS moves the data inside a database. Discovering what exists on-premises before either runs is the job of AWS Application Discovery Service (also now closed to new customers, in favor of AWS Transform).