Pro ยท Lesson 7 of 11
Same engine family can often keep its schema. A different engine needs a conversion step before DMS is asked to load rows into tables that make sense.
Sign in to save your place.
Moving shelves between two shops of the same chain is mostly carrying boxes. Moving into a shop that uses different shelf labels means reprinting the labels first. The boxes come after.
PostgreSQL to PostgreSQL, or PostgreSQL to Aurora PostgreSQL, is homogeneous. Tables, types, and many constraints already match. You still check extensions, large objects, sequences, and users. Native logical replication or dump and restore can be a better tool than DMS. DMS is one option, not a rule.
The PostgreSQL replication lesson covers slots if you choose the database's own replication instead of DMS.
Oracle or SQL Server to Aurora PostgreSQL is heterogeneous. Run a conversion assessment with DMS Schema Conversion or with AWS SCT, fix the action items a person must touch, and apply the schema before the full load. DMS can create tables, and those tables will not include a correct translation of procedural code.
Generative AI conversion, where AWS offers it for a pair, still needs a person to read the result. The user guide marks which pairs have it. Do not assume every source and target pair does.
A virtual target in DMS Schema Conversion lets you assess before the real database exists. You still need a real target before you apply the code and start the data task.
| Path | Schema first? | Then |
|---|---|---|
| Same engine | Usually create or restore the schema you already trust | Full load, CDC, or a native tool |
| Different engine | Yes. Convert and review | Full load and CDC into that schema |
| Data warehouse such as Redshift | Design the target tables for the warehouse | Load. Do not expect transactional procedures to come along |
Step 1 of 3The conversion assessment lists automatic work and manual work.
Step 2 of 3Apply reviewed schema to the target. Read generated code.
Step 3 of 3The DMS data task fills tables that already match the design.
Write homogeneous or heterogeneous, and the tool that will touch the schema. Unknown is allowed if you have not opened the sources-and-targets page.
Source:
Target:
Path:
Schema tool: none, DMS Schema Conversion, or AWS SCT
Action items a person must fix before load:
Data tool: DMS, or a native option we will compare:
Checked against the linked documentation in October 2026. A later release can move a flag, a price, or a supported pair. Use the page for the version you installed.