4.3 - Database migration and refactoring tools (SCT, DMS, Refactor Spaces)
AWS Solutions Architect Professional objective 4.3 covers database migration and application refactoring tooling. A heterogeneous move like Oracle to Aurora PostgreSQL pairs the AWS Schema Conversion Tool (SCT) to convert schema and code with AWS Database Migration Service (DMS) to move the data, and SCT's assessment report tells you up front which objects convert automatically and which need manual rework. AWS Transfer Family exposes managed SFTP, FTPS and FTP endpoints backed by S3 or EFS with custom identity providers, License Included avoids buying and tracking Windows or SQL Server licenses, and you decompose a monolith incrementally with the strangler fig pattern routing migrated paths through a facade. Purpose-built targets modernize access patterns: DynamoDB with DAX for a high-read session store, Amazon DocumentDB for MongoDB compatibility, and AWS Migration Hub Refactor Spaces to front old and new services behind one managed API Gateway. An idle-but-triggered EC2 job refactors to Lambda on S3 events. Expect scenarios about schema conversion, keeping SFTP clients, or incremental decomposition, and ask which tool fits.
Convert schema then move data (different engines) = SCT + DMS. Managed SFTP backed by S3 keeping existing clients = AWS Transfer Family. One managed API Gateway fronting old and new services = Migration Hub Refactor Spaces.
Practice questions
1. A company migrates a self-managed Oracle database to Amazon Aurora PostgreSQL. Which TWO AWS tools together handle the schema conversion and the data movement? (Choose TWO.)
- Amazon Athena to run interactive SQL queries directly against the source database files
- AWS Schema Conversion Tool (SCT) to convert schema and code (correct answer)
- AWS Glue crawler to catalog the Oracle tables only
- AWS Database Migration Service (DMS) to move the data (correct answer)
For a heterogeneous migration (different engines), SCT converts the schema, stored procedures, and code from Oracle to PostgreSQL, and DMS then replicates the actual data. Athena queries S3, and a Glue crawler only builds a catalog; neither converts schema nor migrates a live database.
2. A company needs to copy 90 TB from an on-premises NFS share to Amazon S3 over an existing Direct Connect link, with automatic verification and incremental syncs. Which service is designed for this?
- Amazon S3 Transfer Acceleration over the public endpoint
- AWS DataSync running scheduled online transfer tasks (correct answer)
- AWS Snowball Edge shipped back and forth physically
- Amazon Kinesis Data Firehose streaming the files
DataSync is built for high-speed online transfers from NFS/SMB/HDFS to S3/EFS/FSx over network links, with built-in integrity checks and scheduled incremental syncs. Transfer Acceleration only speeds S3 uploads (no NFS agent), Snowball is offline for low-bandwidth cases, and Firehose streams records, not file shares.
3. A production PostgreSQL database must move to RDS with only a few minutes of downtime, staying in sync while the team validates the target before cutover. Which DMS capability achieves this?
- A one-time full load followed by a long maintenance window
- Exporting a snapshot and restoring it during a weekend outage
- Native PostgreSQL logical replication configured manually end to end
- Full load plus ongoing change data capture (CDC) replication (correct answer)
DMS full load plus CDC copies existing data, then streams ongoing changes so the target stays current; you cut over in a short window once caught up. A one-time load or snapshot restore forces a long outage, and hand-rolled native replication is exactly the manual work DMS removes.
4. Partners currently push data files to an on-premises SFTP server. The company wants to migrate this ingestion to AWS while keeping partners' existing SFTP clients and credentials workflow. Which TWO statements about AWS Transfer Family are correct? (Choose TWO.)
- It requires every external partner to rewrite their client applications to call the S3 object API directly
- It exposes managed SFTP/FTPS/FTP endpoints backed by S3 or EFS (correct answer)
- It supports custom identity providers for existing credential workflows (correct answer)
- It only works if partners install the AWS CLI on their own servers
Transfer Family provides fully managed SFTP/FTPS/FTP endpoints that land files directly in S3 or EFS, and it can use a custom identity provider so existing partner credentials keep working. Partners need not change clients or install the CLI, since they still speak standard SFTP.
5. An engineer has a single custom-built on-premises VM image and simply needs it available as an EC2 AMI, without setting up ongoing replication. Which service does this directly?
- AWS Application Migration Service for continuous block replication
- VM Import/Export to import the image as an EC2 AMI (correct answer)
- AWS Backup restoring the VM from a cross-account vault
- Amazon AppStream 2.0 streaming the desktop to users
VM Import/Export takes an existing VM image (VMDK, VHD, OVA) and imports it as an EC2 AMI for a one-off move, no agents or replication needed. MGN is for continuous, at-scale rehosting, Backup restores AWS resources, and AppStream streams desktop apps.
6. After migrating a web app to AWS, the team wants a cutover that can send a small percentage of live traffic to the new stack first and roll back instantly if errors spike. Which approach supports this?
- Route 53 weighted records shifting traffic gradually between stacks (correct answer)
- A single hard DNS cutover configured with a very long TTL pointing straight at the new stack
- Deleting the old environment before validating the new one
- A single maintenance window that swaps all users at once
Route 53 weighted routing lets you send a small slice of traffic to the new stack, watch metrics, increase gradually, and revert weights instantly on trouble, a safe blue/green cutover. A long-TTL hard change, deleting the old stack early, or an all-at-once swap all remove the ability to roll back quickly.