AWS DMS vs Azure Database Migration Service (2026)

Quick verdict

Short answer

In practice the choice is usually made by the target cloud: AWS DMS when the destination is on AWS, Azure DMS when it is on Azure. Beyond that, choose AWS DMS when you need ongoing CDC replication, heterogeneous moves (for example SQL Server or Oracle to PostgreSQL, with DMS Schema Conversion first) or targets such as Redshift, S3, Kinesis or Kafka. Choose Azure DMS when you are modernising SQL Server into Azure SQL Managed Instance, SQL Server on Azure VMs or Azure SQL Database, and want schemas moved with the data. Note that on Azure the older DMS (classic) for SQL Server and the Azure Data Studio migration extension are both retired, and Azure SQL Database is an offline-only target in DMS.

How we know: This comparison is research-based: scope, supported sources and targets, migration modes, schema handling, retirements and pricing were checked against the AWS DMS User Guide, the AWS DMS pricing page and AWS price data, Microsoft Learn, a Microsoft Q&A retirement notice posted by Microsoft staff, the Microsoft Data Migration blog and the Azure DMS pricing page on 7 November 2026. We have not run either service, so nothing here is a measured result.

AWS Database Migration Service (AWS DMS) copies data from a source to a target through a replication instance, a DMS Serverless replication, or (for like-to-like MySQL, PostgreSQL and MongoDB moves) an instance-free homogeneous data migration. Its source list covers Oracle, SQL Server, MySQL, MariaDB, PostgreSQL, MongoDB, SAP ASE and IBM Db2, plus managed services elsewhere such as Azure SQL Database and Google Cloud SQL. Targets include the same engines and AWS services such as Amazon Redshift, S3, DynamoDB, Kinesis Data Streams, OpenSearch Service, Neptune and Apache Kafka. AWS documents that one endpoint must be on an AWS service. The site's AWS DMS guide walks through a SQL Server to RDS for PostgreSQL migration.

Azure Database Migration Service (Azure DMS) is driven from the Azure portal, PowerShell and the Azure CLI. Microsoft Learn says it "currently supports SQL Database modernization to Azure": SQL Server (including Amazon RDS for SQL Server) to Azure SQL Database, Azure SQL Managed Instance and SQL Server on Azure Virtual Machines. Its scenario status page also lists MySQL to Azure Database for MySQL flexible server, PostgreSQL to Azure Database for PostgreSQL flexible server, MongoDB to Azure Cosmos DB, and Oracle to the Azure SQL targets through SSMA (in preview).

Which Azure DMS? On 7 November 2026 the current product is the Azure DMS reached through the Azure portal, PowerShell and the Azure CLI. DMS (classic) for SQL Server scenarios was retired: new classic SQL resources could not be created from 1 August 2023, and Microsoft staff announced that classic SQL Server scenarios would not be supported after 15 March 2026 (classic MySQL, PostgreSQL and MongoDB scenarios were out of scope of that notice). The Azure SQL Migration extension for Azure Data Studio retired together with Azure Data Studio on 28 February 2026. For the guided SQL Managed Instance path, Microsoft now points to the SQL Server migration experience in Azure Arc.

Side by side

Aspect	AWS DMS	Azure Database Migration Service
Scope	General replication between many engines and AWS data stores; one endpoint must be on AWS	Migration into Azure: mainly SQL Server to Azure SQL, plus MySQL, PostgreSQL and MongoDB scenarios
Main sources	Oracle, SQL Server (not Express), MySQL, MariaDB, PostgreSQL, MongoDB, SAP ASE, Db2, Azure SQL Database, Google Cloud SQL, S3	SQL Server and Amazon RDS for SQL Server; MySQL variants; PostgreSQL; MongoDB; Oracle via SSMA (preview, offline)
Main targets	RDS, Aurora, Redshift, S3, DynamoDB, Kinesis, Kafka, OpenSearch, Neptune, DocumentDB, self-managed engines	Azure SQL Database, Azure SQL Managed Instance, SQL Server on Azure VMs, Azure Database for MySQL and PostgreSQL flexible server, Azure Cosmos DB
Ongoing replication	Full load, full load plus CDC, or CDC only	Online (continuous sync) to SQL Managed Instance, SQL Server on Azure VMs, MySQL, PostgreSQL and Cosmos DB; Azure SQL Database is offline only
Schema handling	Creates tables and primary keys only; DMS Schema Conversion (free) or legacy AWS SCT for the rest	Migrates schemas for the SQL Server scenarios; SSMA for Oracle, Db2, MySQL, Access and SAP ASE to SQL Server
How data moves	Replication instance you size, DMS Serverless (DCUs), or homogeneous data migrations using native tools	SQL Managed Instance and Azure VMs: backup and restore from an SMB share or blob storage; Azure SQL Database: logical copy, with a self-hosted integration runtime as compute
Interfaces	AWS console, AWS CLI, API	Azure portal, PowerShell, Azure CLI; Azure Arc for guided SQL Managed Instance migrations
Not included	Secondary indexes, foreign keys, users and sequences are not carried by the data task	Portal version: no assessment, SKU recommendation, login migration or TDE-encrypted databases
Retired variants	DMS Fleet Advisor ended support on 20 May 2026; AWS calls SCT legacy	DMS (classic) SQL Server scenarios (15 March 2026); Azure Data Studio extension (28 February 2026)
Pricing model	Per hour per replication instance, per DCU-hour (Serverless) or per migration hour; storage and transfer extra	Standard tier free (offline); Premium tier for online migrations, free for 183 days then hourly; the pricing page labels these tiers DMS (classic)

Key differences

Replication engine versus migration path into one cloud

AWS DMS is closer to a general replication tool. You define a source and a target endpoint, pick a migration type (full-load, full-load-and-cdc or cdc) and let a task run for as long as you need, which also makes it usable for keeping a copy in sync or feeding change streams into Kinesis or Kafka. The constraint is that one of the two endpoints must be on an AWS service; AWS documents that DMS cannot move data from one on-premises database to another.

Azure DMS is organised around migration scenarios with a cutover at the end. For SQL Server, an online migration keeps the target in sync until you cut over, but only for Azure SQL Managed Instance and SQL Server on Azure VMs; to Azure SQL Database the scenario status page lists offline migration only. Microsoft's February 2026 blog on Azure Data Studio alternatives names Striim, a third-party product, for online moves to Azure SQL Database. In our view, that difference matters most for large SQL Server databases bound for Azure SQL Database, where an offline window may be hard to arrange.

What the two services cover

AWS DMS reads from Oracle 10.2 to 19c, SQL Server 2012 to 2022 (2008 and 2008 R2 on DMS 3.5.4; Express is not supported), MySQL 5.5 to 8.4, MariaDB, PostgreSQL 9.4 to 18, MongoDB 3.x to 8.0, SAP ASE and IBM Db2, and from managed services on other clouds such as Azure SQL Database, Azure Database for MySQL and PostgreSQL flexible server and Google Cloud SQL. Targets go beyond relational engines to Redshift, S3, DynamoDB, OpenSearch, Neptune, ElastiCache, Kinesis and Kafka.

Azure DMS is narrower and Azure-bound. For SQL Server it supports on-premises SQL Server and Amazon RDS for SQL Server as sources, generally available, and Oracle through SQL Server Migration Assistant in preview. For open source engines, Microsoft lists MySQL sources (including Amazon RDS, Aurora, Google Cloud SQL and Percona) to Azure Database for MySQL flexible server, PostgreSQL and Amazon RDS for PostgreSQL to Azure Database for PostgreSQL flexible server (online only), and MongoDB to Azure Cosmos DB. The DMS PostgreSQL tutorials are labelled DMS (classic), and Microsoft recommends the migration service built into Azure Database for PostgreSQL for new PostgreSQL migrations; MongoDB to Azure DocumentDB uses a VS Code extension.

Schema conversion and what each leaves to you

AWS DMS separates data from schema. The data task creates tables and primary keys if they do not exist, but secondary indexes, foreign keys, views, procedures, users and sequences come from DMS Schema Conversion (a free, web-based feature built on the AWS SCT engine that can use generative AI for objects its rules cannot convert), the legacy AWS Schema Conversion Tool, or your own scripts. That split is what makes heterogeneous moves possible: convert and apply the schema first, then run the data task with the target table preparation mode set to do nothing or truncate.

Azure DMS migrates schemas as well as data in its SQL Server scenarios, which suits like-to-like moves into the SQL Server family. Microsoft's comparison table for the portal version lists assessment, SKU recommendations, login migration and TDE-encrypted databases as not supported; Microsoft points to Azure Migrate and Azure Arc for discovery and assessment, and to SSMA for converting Oracle, Db2, MySQL, Access and SAP ASE schemas to SQL Server. Before cutover on either platform, check logins, jobs, sequences or identity values and anything else the tool does not carry; see the database migration guide.

-- T-SQL (SQL Server source): check the recovery model and CDC state before an online migration
SELECT name, recovery_model_desc, is_cdc_enabled
FROM sys.databases
WHERE name = N'mydb';

Retired and replaced variants on Azure

Tutorials for Azure DMS age quickly because the product has changed shape several times. DMS (classic) was the original service with Standard and Premium tiers; for SQL Server scenarios, new classic resources could not be created from 1 August 2023 and support ended after 15 March 2026, according to the retirement notice Microsoft staff posted on Microsoft Q&A. Classic scenarios for MySQL, PostgreSQL and MongoDB were outside that notice. The Azure SQL Migration extension for Azure Data Studio was deprecated and retired with Azure Data Studio on 28 February 2026, according to the Microsoft Data Migration blog.

For SQL Server targets Microsoft now recommends: the SQL Server migration experience in Azure Arc for guided migrations to Azure SQL Managed Instance; the Azure DMS portal for SQL Server on Azure VMs (online or offline) and for offline moves to Azure SQL Database; and the DMS PowerShell cmdlets or Azure CLI for automation at scale on all three. On the AWS side the changes are smaller: DMS Fleet Advisor reached end of support on 20 May 2026, and AWS now calls the downloadable AWS SCT the legacy option next to DMS Schema Conversion.

Running and paying for the migration

AWS DMS gives you sizing decisions: a replication instance class (T3 for tests, C classes for compute, R classes for heavy CDC), Multi-AZ for CDC, allocated storage, VPC and subnet group. DMS Serverless replaces the instance with a capacity range in DMS capacity units (one DCU equals 2 GB of RAM). Cost follows those choices, hour by hour, which makes a long CDC period visible on the bill.

Azure DMS works differently by target. Microsoft's FAQ says migrations to Azure SQL Managed Instance and SQL Server on Azure VMs are physical (backup and restore from an SMB network share or an Azure Blob container), while migrations to Azure SQL Database are a logical copy of table data; when backups are copied from a local file share, or the target is Azure SQL Database, DMS uses a self-hosted integration runtime node as its compute, so that machine is the one to size and monitor. The same FAQ says an online migration requires a DMS instance on the Premium pricing tier. The Azure pricing page lists a free Standard tier for offline migrations and an hourly Premium tier for online migrations, free for 183 days from service creation, but it labels these tiers DMS (classic), so confirm with Microsoft how they apply to the current version before budgeting.

Pricing and licensing

AWS DMS (US East, N. Virginia, USD, excluding tax, from AWS price data published 11 September 2026 and current on 7 November 2026): dms.t3.medium 0.0745 per hour Single-AZ or 0.149 per hour Multi-AZ; dms.r6i.large 0.176 or 0.353 per hour; replication storage 0.115 or 0.23 per GB-month above the included 50 GB (T2 and T3) or 100 GB (C4, R4, R5); DMS Serverless about 0.0819 per DCU-hour Single-AZ or 0.1639 Multi-AZ; homogeneous data migrations 0.09 per migration hour. DMS Schema Conversion is free apart from Amazon S3 storage, data transfer into DMS is free, and T3 CPU credits above the baseline cost 0.075 per vCPU-hour. Database Savings Plans are a third pricing option. The AWS DMS guide has a worked one-month example.

Azure DMS. The Azure pricing page on 7 November 2026 described Azure Database Migration Service (classic): the Standard tier (1, 2 or 4 vCores) for offline migrations, "generally available and free to customers"; the Premium tier (4 vCores) for offline and online migrations, billed hourly, free for 183 days from service creation, with each classic service usable for up to one year and 10 classic services per subscription. The Premium hourly rate was not shown in the page text (it varies by region in the price selector). Microsoft's DMS FAQ says online migrations require a Premium tier instance and links to this page, but the page itself describes the tiers as DMS (classic), so confirm with Microsoft how they apply to the current version. Target resources (Azure SQL, Managed Instance, VMs), the storage that holds backups and the machine running the self-hosted integration runtime are billed at their own rates.

Pricing checked on the vendors' official pages on 7 November 2026. Prices change; confirm before buying.

Where each one leads

AWS DMS strengths

Many sources and targets, including non-relational AWS services such as Redshift, S3, Kinesis and Kafka
Full load, full load plus CDC, or CDC only, for migrations or ongoing replication
DMS Schema Conversion is free and handles heterogeneous schema conversion before the data task
Published hourly prices per instance class, DCU-hour or migration hour
Serverless option that scales capacity between limits you set

Azure Database Migration Service strengths

Built for SQL Server modernisation into Azure SQL Database, SQL Managed Instance and SQL Server on Azure VMs
Migrates schemas with the data in SQL Server scenarios
Online migrations with minimal downtime to SQL Managed Instance and SQL Server on Azure VMs
Backup-and-restore migrations to Managed Instance and Azure VMs, with automation through PowerShell and Azure CLI
Also covers MySQL, PostgreSQL and MongoDB moves into Azure services

Limitations

AWS DMS limitations

One endpoint must be on AWS; on-premises to on-premises moves are not supported
Secondary indexes, foreign keys, users, code objects and sequences need Schema Conversion or scripts
SQL Server Express is not supported as a source
You size and pay for replication instances by the hour, including during long CDC periods
LOB settings trade speed against truncation and need planning

Azure Database Migration Service limitations

Azure-bound: targets are Azure services only
Azure SQL Database is an offline-only target
Portal version has no assessment, SKU recommendation, login migration or TDE support
Older variants (DMS classic for SQL Server, the Azure Data Studio extension) are retired, so many tutorials are out of date
Pricing page describes its tiers as DMS (classic) and shows no Premium hourly rate in the page text

When to choose each

Choose AWS DMS if

Your target is on AWS: RDS, Aurora, Redshift, S3 or DynamoDB
You need CDC to keep a target in sync for weeks or to feed Kinesis or Kafka
You are changing engines, for example SQL Server or Oracle to PostgreSQL
You want to read from a managed database on another cloud, such as Azure SQL Database or Google Cloud SQL

Choose Azure Database Migration Service if

You are moving SQL Server into Azure SQL Managed Instance or SQL Server on Azure VMs
You want schema and data moved together for a like-to-like SQL Server migration
You need an online cutover into Managed Instance or an Azure VM with minimal downtime
You are migrating MySQL or MongoDB into Azure Database for MySQL or Azure Cosmos DB

When neither is right

A small database that can be stopped for an hour: a native backup and restore or dump tool is simpler; see the database migration guide.
Converting Oracle, Db2, MySQL, Access or SAP ASE schemas to SQL Server: see the SQL Server Migration Assistant guide.
Moving schema changes between environments rather than migrating a database: use a migration or compare tool; see Flyway vs Liquibase and the database comparison tools.
Continuous pipelines from many SaaS and database sources into a warehouse: an ELT tool fits better; see Airbyte vs Fivetran.

Final recommendation

Bottom line

For most teams the target cloud decides: AWS DMS for moves onto AWS, Azure DMS for moves into Azure. Where both are possible, AWS DMS is the more flexible replication engine, with CDC, heterogeneous moves through free Schema Conversion and many non-relational targets, at the cost of sizing instances and handling indexes, users and sequences yourself. Azure DMS is the more direct route for SQL Server into Azure SQL Managed Instance or Azure VMs, moving schema and data together with an online option, but it is offline-only to Azure SQL Database and its older variants were retired in 2026. Check that every tutorial you follow describes the current product, and confirm Azure DMS costs with Microsoft.

Frequently asked questions

Is Azure Database Migration Service retired?

No. The current Azure DMS, used through the Azure portal, PowerShell and the Azure CLI, is still documented on Microsoft Learn. What was retired is DMS (classic) for SQL Server scenarios (no new resources from 1 August 2023, not supported after 15 March 2026) and the Azure SQL Migration extension for Azure Data Studio (retired with Azure Data Studio on 28 February 2026).

Can Azure DMS do online migrations to Azure SQL Database?

Not according to Microsoft's scenario status page checked on 7 November 2026: SQL Server to Azure SQL Database is offline only, while online migrations are supported to Azure SQL Managed Instance and SQL Server on Azure VMs.

Can AWS DMS migrate a database to Azure?

Not as a general rule. AWS documents that one endpoint must be on an AWS service. AWS DMS can read from Azure SQL Database and Azure Database for MySQL or PostgreSQL flexible server as sources, but its targets are AWS services or databases reachable from AWS.

Does either service convert stored procedures?

AWS DMS does not in the data task; DMS Schema Conversion (free) or the legacy AWS SCT converts schemas and code for heterogeneous moves. Azure DMS migrates schemas for SQL Server to SQL Server-family targets; for other engines Microsoft points to SQL Server Migration Assistant.

How much does each cost?

AWS DMS publishes hourly prices per replication instance, DCU-hour and migration hour (for example 0.149 USD an hour for a Multi-AZ dms.t3.medium in US East). Azure DMS offline migrations use the free Standard tier; Microsoft's FAQ says online migrations need the Premium tier, which the pricing page lists as free for 183 days and then hourly, with the tiers labelled DMS (classic) and no hourly rate in the page text when checked.

Sources

Checked November 2026.

How we research comparisons: our editorial method.