Major AWS Outage Disrupts Global Internet Services
A massive Amazon Web Services (AWS) outage on October 20-21, 2025, disrupted digital services for over 2,000 companies worldwide. The 15-hour incident, centered in AWS’s largest US-East-1 data center cluster, affected major platforms including Amazon, Snapchat, Signal, ChatGPT, Perplexity, Canva, and Epic Games before being fully resolved at 6:53 PM ET.
Key Takeaways
- 15-hour AWS outage affected 2,000+ companies globally
- DNS error in DynamoDB service caused widespread disruption
- US-East-1 region, AWS’s oldest and largest, was epicenter
- Amazon implementing additional protections to prevent future incidents
What Caused the AWS Outage?
The disruption originated from a DNS error affecting DynamoDB database APIs in the US-East-1 region. The Domain Name System (DNS) translates web URLs into IP addresses, and when it fails, services cannot receive incoming traffic even if operational.
AWS advised affected users to flush their DNS caches to resolve DynamoDB service endpoint issues. Cybersecurity experts confirmed this was an “availability” issue where the system couldn’t correctly decide which server to connect to, creating a ripple effect.
Why US-East-1 Is Critical
Built in 2006, US-East-1 is AWS’s first and most active region, serving as the default for many services. Its global infrastructure means even European services can be affected by outages through services like DynamoDB Global Tables.
History of AWS Outages
US-East-1 has a history of major disruptions. The December 2021 outage, considered AWS’s most severe, lasted seven hours and cost S&P companies $150 million due to a debugging command typo. Despite architectural improvements, another significant outage occurred in June 2023, affecting 100 services for four hours.
Future Outlook and Expert Analysis
Experts warn that cloud outages may increase as AI adoption drives more data to cloud platforms. The incident highlights the risks of over-reliance on dominant cloud providers.
Gergely Orosz, author of ‘The Pragmatic Engineer,’ noted unexpected casualties including Postman and Eight Sleep, stating: “In both cases, things should have worked locally! But clearly the dev teams found it simpler to take on a cloud dependency – and made no prep for an AWS region outage.”
Amazon’s Response
Amazon will temporarily disable DynamoDB DNS Planner and add additional protections to prevent DNS errors. The AWS team is also improving internal testing to identify similar problems that prolonged the outage.
