The AWS us-east-1 Outage of October 2025: DNS, DynamoDB and a Day of Downtime

Retrospective: this article looks back at events from October 2025, written in 2026 with the benefit of hindsight.

On October 20, 2025, AWS's US-EAST-1 region suffered a major outage that disrupted a wide range of services — from banking and gaming apps to smart home devices, airline systems and productivity tools — for much of the day.

What happened

According to AWS's post-event summary, the trigger was a latent race condition in the automated DNS management system for Amazon DynamoDB in US-EAST-1. The automation left an empty DNS record for DynamoDB's regional endpoint, making it unreachable. Because many AWS services and customer applications depend on DynamoDB, failures cascaded — including to EC2 instance launches and network load balancer health checks — extending recovery even after the DNS issue was corrected. Full recovery across services took many hours.

Why it mattered

• Concentration risk: US-EAST-1 is AWS's oldest and largest region, and many global services and control-plane functions depend on it.
• Hidden dependencies: customers discovered dependencies on DynamoDB and US-EAST-1 they hadn't documented — sometimes through third-party SaaS providers.
• Cascading failures: recovery of one component triggered load and delays in others.

Lessons in hindsight

• Map critical dependencies, including those of your SaaS providers.
• Design for regional failure for the few services that justify it.
• Static stability: applications should keep running without control-plane changes during an outage.
• Practice failover — and know which DNS, identity and data services must be available in the secondary region.

In hindsight

The 2025 outage echoed 2017 and 2021. Each time, the question for leadership was the same: which services truly need multi-region resilience, and have we tested it? The answer is rarely "everything" — but it's rarely "nothing."