Read summarized version with
About project
Working time:
2024 to 2025
Industry:
Cash logistics
The service:
Overview
The customer is a North American cash logistics company that moves and processes cash for banks and retailers. Its software platform schedules routes, tracks vault inventory and gives customers a portal for orders and reporting.
The Challenge
The platform had grown across two environments: core services in Microsoft Azure and older systems in a company data centre. Running both meant two sets of tools, skills and bills.
1. A Split Platform
Applications in Azure depended on databases and file shares in the data centre, adding latency and failure points.
2. Hand-Built Environments
Test and production were built by hand, so they drifted apart.
3. Downtime Risk
Customers expected the portal and route planning to run every day of the year.
4. Two Environments to Pay For
Licenses, support and staff time were split across both environments.
Technology & Approach
We planned the move in waves, each with a downtime window and a rollback, and built every AWS environment as code before any data moved.
Implementation
Phase 1: Assessment and Wave Plan
Inventoried services, data stores, identity, networking and pipelines in both environments and mapped each to an AWS target.
Phase 2: Landing Zone and Code
Built a landing zone with AWS Control Tower and defined every environment in Terraform.
Phase 3: Migration Waves
Moved containerized services to Amazon EKS, databases with AWS Database Migration Service to Amazon RDS for SQL Server and Amazon Aurora PostgreSQL, file shares with AWS DataSync, and remaining servers with AWS Application Migration Service.
Phase 4: Cutover and Hypercare
Rebuilt the pipelines, switched DNS in planned windows with a tested rollback, then ran hypercare and cost tuning.
Customization & Client Adaptation
Cash operations run every day, so the plan was shaped around the business calendar.
Sign-In Unchanged for Staff
Staff kept signing in with their existing directory through AWS IAM Identity Center.
Validated Data Moves
Row counts and checksums were compared before each cutover, and the old system stayed read-only until sign-off.
Low-Volume Windows
Cutovers ran in windows agreed with operations, away from peak collection days.
Security from Day One
Encryption, logging and guardrails were part of the landing zone, not added after the move.
If you find this case interesting, we recommend taking a closer look at
The Results
The platform now runs entirely on AWS. Every migration wave cut over in its planned window, and the company retired its data centre and Azure subscriptions after the final wave.
Long-Term Value
- Every environment reproducible from code.
- One cloud to run, secure and pay for.
- A foundation for analytics on route and vault data.
Business Impact
- No customer-facing outage during the moves.
- Test and production environments now match.
- Lower latency between applications and their data.
