Multi-region disaster recovery for Amazon API Gateway WebSocket APIs

This article presents a multi-region disaster recovery pattern for Amazon API Gateway WebSocket APIs that automatically reconnects clients and restores state during regional failover.

• Active-passive architecture with primary and secondary regions sharing state via DynamoDB Global Tables
• Route 53 failover routing detects regional failures and redirects traffic automatically
• Canary health probe opens WebSocket connections every minute to detect failures independently
• CloudWatch alarms trigger Route 53 health check updates after two consecutive probe failures
• Client-side exponential backoff with jitter prevents thundering herd during reconnection
• Clients restore application state by sending getState message after reconnecting to secondary region
• Heartbeat mechanism (30-second interval, 35-second timeout) monitors connection health
• Demo auction application shows real-time state synchronization across regions during failover

This pattern enables seamless failover for real-time applications like trading platforms, collaborative editors, and gaming leaderboards by separating application state from WebSocket connections.