Use Case · Financial Services · Database Migration
How Itaú Unibanco migrated 400+ SQL Server
databases to AWS at enterprise scale
Latin America's largest bank modernized its entire Microsoft SQL Server estate onto Amazon RDS — eliminating on-premise licensing debt, improving resilience, and unlocking cloud-native scalability across its core banking platform.
The challenge
As Latin America's largest financial institution, Itaú Unibanco operates at extraordinary scale — serving over 60 million customers across Brazil and a further 20 countries. Its core banking infrastructure relied on a vast estate of Microsoft SQL Server databases, spread across on-premise data centers with significant licensing overhead, limited elasticity, and growing operational complexity.
The mandate was clear: migrate all 400+ SQL Server databases to AWS with minimal disruption to 24/7 banking operations, maintain strict compliance with Brazil's LGPD data privacy law and BACEN regulatory requirements, and position the bank's data platform for long-term cloud-native growth.
Migration approach
Homogeneous — SQL Server to RDS SQL Server
The majority of databases were migrated on a like-for-like basis using AWS DMS and native SQL Server backup/restore, retaining existing schema, stored procedures, and application compatibility with zero code changes required.
Primary path for mission-critical transactional databases requiring fastest cutover
Heterogeneous — SQL Server to Aurora PostgreSQL
Selected workloads were replatformed from SQL Server to Amazon Aurora PostgreSQL using AWS SCT for schema conversion and custom remediation scripts, enabling open-source cost savings and deeper AWS-native integration.
Applied to analytically intensive workloads to maximize long-term TCO reduction
Lift & Shift — On-premise to EC2 (Interim)
Complex databases with deep OS-level dependencies were first moved to EC2-hosted SQL Server as an interim step, decoupling data center exit from application modernization timelines.
Used for highest-complexity systems requiring multi-phase migration
Decommission & Consolidation
Legacy databases with redundant or inactive workloads were identified, archived to Amazon S3, and formally decommissioned — eliminating unnecessary licensing and operational overhead.
Resulted in a leaner, consolidated post-migration database estate
Near-zero downtime methodology
With 24/7 banking operations running across 60+ million customers, any unplanned disruption was unacceptable. A rigorous cutover framework was engineered for every migration wave:
Parallel run & CDC
Source and target environments ran simultaneously with AWS DMS Change Data Capture maintaining sub-second replication lag ahead of each cutover window.
Automated validation gates
Row-count checks, checksum verification, and 150,000+ SQL regression tests required a 100% pass rate before any cutover could proceed.
Rollback-ready cutovers
Automated runbooks executed each cutover with rollback capability retained for 7 days, and a dedicated hypercare war room active for 72 hours post-event.
Results
Ready to modernize your database estate at scale?
Talk to our team →