How We Would Rescue a Payments Platform from Single-Region Fragility

02 — Our approach

Our approach would start with an active-active multi-region setup across eu-west-1 and eu-central-1, but not the way textbooks recommend. Their stateful payment ledger could not be split without a 6-month database refactoring project they could not afford. So we would run the ledger in eu-west-1 with read replicas in eu-central-1, and route traffic through a custom failover layer that degrades gracefully — reads stay fast, writes queue and replay after recovery. We would use Terraform for infrastructure, GitOps for deployments, and write a detailed runbook for the failover logic because the team needs to understand it, not just trust it. The compliance piece would be harder than the infrastructure: SOC 2 auditors want evidence of every change, so we would wire immutable audit trails into every deployment and access decision.