Amazon Web Services back online after 15-hour global outage takes down thousands of websites

Amazon Web Services has announced that all cloud operations have returned to normal following a massive 15-hour outage that crippled over 1,000 websites and applications worldwide, affecting millions of users across banking, gaming, and social media platforms for about entire day.

The disruption, which began at approximately 3 a.m. ET on October 20, left millions unable to access essential services ranging from banking apps and social media platforms to gaming sites and smart home devices. AWS confirmed full restoration by 6 p.m. ET, but not before the outage wreaked havoc across the digital landscape, affecting everything from Snapchat streaks to Premier League technology.Among the casualties were major platforms including Duolingo, Roblox, Zoom, Coinbase, and financial institutions like Lloyds Bank, Halifax, and Bank of Scotland. Even Amazon's own services suffered, with Ring doorbell cameras going dark, Alexa assistants falling silent, and the company's e-commerce platform displaying error messages with pictures of dogs apologizing to frustrated shoppers.

DNS failure in Virginia data center sparked internet-wide chaos

The culprit was a DNS resolution failure affecting DynamoDB, AWS's critical database service, in the company's US-EAST-1 region—its largest and oldest data center located in northern Virginia. The Domain Name System, which functions as the internet's address book by converting website names into numeric IP addresses, essentially suffered temporary amnesia."When the system couldn't correctly resolve which server to connect to, cascading failures took down services across the internet," explained Davi Ottenheimer, vice president at data infrastructure company Inrupt.

IT security experts noted that when domain name resolution stops working, even well-designed applications and services can come to a complete halt.AWS acknowledged the problem stemmed from "an underlying internal subsystem responsible for monitoring the health of our network load balancers," which distributes traffic across various servers. The company implemented multiple mitigation strategies throughout the day, though some services experienced additional waves of disruption as engineers worked to address not just symptoms but root causes.

Third outage in five years raises alarm over infrastructure dependency

This marks the third major internet outage in five years originating from AWS's northern Virginia facility, intensifying concerns about the concentration of internet infrastructure. AWS commands approximately 30% of the global cloud computing market, meaning disruptions at a single provider can cripple vast portions of the digital economy."Once you have a concentrated supply in a handful of monopoly providers, when something like this falls over, it takes a huge percentage of the economy out with it," said Cori Crider, head of the Future of Technology Institute.

Computer science experts from Cornell University and Notre Dame pointed out that companies using AWS haven't adequately invested in backup systems and disaster recovery infrastructure.Downdetector reported receiving 6.5 million user complaints globally, with some services like Grok, Lyft, Claude AI, Hulu, and Reddit continuing to show problems even after AWS declared the underlying issue resolved. The outage disrupted work for Roblox game developers, left ride-share users stranded, caused Canvas educational software failures affecting university students, and even prevented access to HMRC government services in the UK.

Legal battles and accountability questions loom

The blame game has begun, with legal experts suggesting affected businesses may pursue compensation. The precedent exists: Delta Airlines continues battling CrowdStrike over $500 million in losses from last year's massive outage, which required manually resetting 40,000 servers.Consumer advocates warned of potential scam attempts exploiting the chaos, urging users to avoid unsolicited calls offering solutions. AWS has promised a detailed post-event summary, though such reports typically take weeks or months to complete.