American Express successfully migrated its payments network twice without any customer-impacting downtime, a feat achieved through careful engineering and phased deployment. The migration involved moving live card authorization traffic from a legacy platform to a new microservices-based architecture while maintaining strict operational constraints including zero downtime and consistent latency.

In 2018, American Express began modernizing its payments network, a mission-critical distributed system handling global card authorization traffic. The network processes millions of transactions between merchants, acquirers, and issuers, requiring continuous availability and low latency. The migration had to maintain existing payment processing logic, keep throughput consistent, and ensure no transactions were dropped or delayed.

The team employed a three-stage strategy: connection migration, shadow traffic testing, and canary routing. They first introduced a Global Transaction Router (GTR) to handle connections without processing logic, then deployed shadow traffic to validate the new platform against live production data, and finally used canary deployments to gradually route traffic to the new system. This approach allowed them to migrate without big-bang cutovers or duplicated development effort.