On October 21, 2025, a single AWS region US-EAST-1 in Northern Virginia went down, shaking the global internet. Apps and platforms like Netflix, Vercel, Reddit, PlayStation, Roblox, Fortnite, Robinhood, Coinbase, Venmo, Snapchat, Disney+, DoorDash, and even Amazon.com all went dark.
The Root Cause
A misconfigured DNS setting inside a subsystem responsible for resolving AWS service endpoints especially Amazon DynamoDB broke the chain. DNS acts as the internet’s address book. When it failed, apps couldn’t locate their databases or APIs, turning them into instant vaporware.
The Ripple Effect
Though AWS fixed the issue within a couple of hours, a massive backlog of Lambda invocations, SQS messages, and API retries followed. This queue storm extended downtime for hours, proving how even short failures can echo across the cloud.
Why US-EAST-1 Matters
US-EAST-1 isn’t just another region it’s AWS’s oldest and busiest, hosting a large share of global traffic. Despite its multiple availability zones, a single misstep inside one system triggered a failure visible worldwide.
The Bigger Lesson
The outage exposed a deeper issue: centralization risk. So much of the internet depends on one provider and often, one region. Cloud computing promises resilience, but true reliability needs distribution, diversity, and independence.
Final Thought
The October 21 outage was a wake-up call for engineers and executives alike. A tiny DNS tweak broke the world’s biggest cloud. It’s time to rethink resilience because when US-EAST-1 sneezes, the internet catches a cold.