Digital Resilience: Failover-Resilient Serverless Architecture

September 1, 2026

Context

Built independently for the EDN Hackathon “Future Builders” (Escola da Nuvem × AWS), Case B: Digital Resilience — a challenge to design infrastructure that stays available under regional failure conditions.

Architecture

• AWS WAF + CloudFront in front of a dual-origin setup: S3 primary in sa-east-1, secondary failover origin in us-east-1
• API Gateway with direct-to-SQS integration — decoupling ingest from processing so a backlog doesn’t become an outage
• Lambda worker consuming from SQS, writing to DynamoDB
• CloudWatch + SNS for observability and alerting across the failover path

What I actually tested

This wasn’t a diagram exercise. I validated the architecture end-to-end with real form submissions and DynamoDB scans — confirming data landed correctly under both normal and failover conditions. I also audited a teammate’s Terraform-provisioned environment against the project requirements, which surfaced gaps a diagram review alone wouldn’t have caught.

Why this design

Coming from 25 years of carrier-grade uptime requirements, the instinct here wasn’t “make it work” — it was “make it survive losing a region without anyone noticing.” The dual-origin failover and SQS decoupling are the same pattern I’d apply to any system where downtime has a real cost, just expressed in AWS-native primitives instead of telecom infrastructure.

Code and architecture diagrams: GitHub