Automation glitch triggered cascading failure
AWS published its official postmortem for the October 2025 outage, attributing the disruption to a misconfigured automated scaling policy that overrode safety limits on load balancers. The change drained capacity in multiple regions and left services unable to route traffic, knocking major streaming, fintech, and e-commerce customers offline for hours. Amazon says new guardrails now require multi-factor approvals before similar automation changes ship to production.
How engineers restored service
Response teams isolated the problematic scaling stack, rotated DNS entries, and rerouted traffic through unaffected regions within four hours. Amazon deployed temporary compute bursts to absorb backlog traffic while rebuilding the impacted control plane. The company added validation layers that simulate failure scenarios and expanded telemetry hooks for customers using control tower dashboards. Our multicloud pipeline guide outlines similar chaos testing routines teams can borrow. Design resilient runbooks
Next steps for cloud leaders
Technology leaders should review their own automated scaling policies, ensuring change management includes peer review and fast rollback paths. Update continuity plans that assume a single region failure, and verify DNS, IAM, and observability tooling can survive a multi-region incident. Compare AWS’s new controls with redundancy strategies from other providers covered in our cloud outage playbook. Evaluate AWS’s evolving roadmap