Module 03
Migrate databases with zero downtime using AWS DMS, and move large-scale data with DataSync, Transfer Family, and Storage Gateway.
AWS DMS enables you to migrate databases to AWS with minimal downtime. The source database remains fully operational during migration, making it ideal for production environments where availability is critical.
For heterogeneous migrations (different database engines), AWS SCT converts source schemas to the target format. It handles table definitions, stored procedures, and application SQL code. SCT is now built into DMS for many common database type combinations.
A fully-managed capability that automates migration planning by discovering and analyzing your on-premises database fleet. It assesses servers and recommends migration paths before you begin the actual migration work.
For AnyCompany Bank, zero-downtime migration is non-negotiable for transaction processing databases. DMS with CDC ensures that payment records, account balances, and transaction histories remain synchronized between source and target until the final cutover — typically scheduled during low-traffic windows.
The migration pattern you choose depends on your target architecture goals, acceptable complexity, and whether you're changing database engines.
Move the database as-is onto an EC2 instance. The database engine stays the same, you manage it yourself, but it runs on AWS infrastructure. Minimal changes required.
Same database engine, different platform. For example, moving an on-premises Oracle database to Amazon RDS for Oracle. You gain managed service benefits (automated backups, patching, HA) without changing the engine.
Different database engine entirely. For example, migrating from Oracle or SQL Server to Amazon Aurora or PostgreSQL on RDS. Requires schema conversion (via SCT) but can deliver significant licensing cost savings.
Amazon RDS is a managed relational database service supporting MySQL, Oracle, SQL Server, MariaDB, PostgreSQL, Db2, and Aurora. It handles routine administration tasks — hardware provisioning, patching, backups, and failover — so teams can focus on application logic rather than database operations.
AnyCompany Bank's heterogeneous migration from Oracle to Aurora PostgreSQL can eliminate millions in annual licensing costs. However, the schema conversion requires careful testing of stored procedures that implement business logic for interest calculations, fee processing, and regulatory reporting.
Beyond databases, organizations need to move files, archives, and unstructured data. AWS provides three primary services for different data movement patterns.
An online transfer service that simplifies and accelerates moving data between on-premises storage systems and AWS. It supports NFS, SMB, self-managed object storage, HDFS, and can transfer to S3, EFS, or FSx.
Provides managed file transfer using standard protocols (SFTP, FTPS, FTP, AS2) directly into S3 or EFS. Integrates with existing identity providers and eliminates the need to manage FTP server infrastructure.
Provides hybrid cloud storage with local caching, bridging on-premises applications with cloud storage backends. Three gateway types serve different workload patterns:
| Gateway Type | Protocol | Backend Storage | Use Case |
|---|---|---|---|
| File Gateway | NFS / SMB | Amazon S3 | File-based applications needing cloud-backed storage with local cache |
| Volume Gateway | iSCSI | S3 + EBS Snapshots | Block storage with cloud backup and disaster recovery |
| Tape Gateway | iSCSI VTL | S3 + S3 Glacier | Replace physical tape infrastructure with cloud-backed virtual tapes |
AnyCompany Bank uses Transfer Family for SWIFT file processing — payment instructions arrive via SFTP and need to land directly in S3 for downstream processing. Storage Gateway's Tape Gateway replaces aging tape libraries used for regulatory archive retention (7+ year requirements for transaction records).
The feasibility of online data migration depends on two factors: how much data you need to move, and how much network bandwidth is available. This calculator helps determine whether online transfer is practical for your timeline.
| Data Volume | 100 Mbps | 1 Gbps | 10 Gbps |
|---|---|---|---|
| 1 TB | ~30 hours | ~3 hours | ~18 minutes |
| 10 TB | ~12 days | ~30 hours | ~3 hours |
| 100 TB | ~124 days | ~12 days | ~30 hours |
| 1 PB | ~3 years | ~124 days | ~12 days |
| 10 PB | ~34 years | ~3 years | ~124 days |
These estimates assume approximately 25% network overhead. The formula: Days = Total Bytes ÷ (Mbps × 125,000 × 86,400). For large-scale transfers exceeding available bandwidth windows, consider AWS Direct Connect for dedicated network capacity.
| Service | Best For | Transfer Method |
|---|---|---|
| DataSync | Bulk online transfers from NFS/SMB/HDFS to S3/EFS/FSx | Online (agent-based) |
| Transfer Family | Protocol-based file transfers (SFTP/FTPS/FTP) into S3/EFS | Online (managed server) |
| Storage Gateway | Hybrid storage with local caching for ongoing access | Online (appliance-based) |
AnyCompany Bank's data migration must account for regulatory requirements around data residency. Transaction archives may need to remain in specific geographic regions, and transfer windows must avoid peak trading hours when network bandwidth is reserved for real-time payment processing.