American Express: Cell-Based Architecture for Resilient Payment Systems
6 hours ago
- #resiliency
- #payment systems
- #cell-based architecture
- American Express's payment platform uses a cell-based architecture for high availability, low latency, and resiliency.
- Cells are independent units containing microservices and data, limiting failure impact and enabling isolated recovery.
- Data locality is achieved through static data replication and deterministic routing via the Global Transaction Router.
- Cross-cell communication is restricted; the Global Transaction Router enforces boundaries and handles rerouting.
- Failures are contained within a cell; transactions are rerouted and restarted in healthy cells without cross-cell state.
- Edge dependencies, like logging and configuration, are managed asynchronously to keep the critical path resilient.
- The architecture allows cells to be taken out of rotation gracefully, supporting maintenance and incident response.