High availability is table stakes in modern serverless systems, but true disaster recovery requires intentional design choices driven by business requirements. This session presents a pragmatic approach to disaster recovery for AWS serverless architectures, focusing on measurable outcomes such as RTO and RPO rather than defaulting to multi-Region complexity.

We’ll examine resilience patterns within a single Region, including retry strategies, asynchronous workflows, idempotency, and failure isolation. From there, we’ll analyze multi-Region disaster recovery models for serverless workloads, comparing active/passive and active/active architectures and their implications on data consistency, traffic routing, and observability.

This talk equips architects and senior engineers with a decision-making framework to design disaster recovery strategies that are cost-aware, operationally realistic, and aligned with business risk.