Replatforming without downtime
- Constraint
- An inherited cloud native product had been designed for speed of development. As the product grew, it became expensive to run and hard to evolve.
- Trade-off
- The old and new systems had to coexist until every path was proven.
- Decision
- I redesigned the platform around managed containers and data services, then migrated traffic route by route using staged routing, load balancing, and blue/green deployment patterns.
- Outcome
- Monthly cloud spend fell by more than 60%, with no migration downtime. Reusable infrastructure patterns made later services faster and safer to stand up. Development velocity was improved through modern CI/CD practices.