ASHBURN, Virginia — Before sunrise on Monday, the internet’s backbone buckled with an AWS outage. Amazon Web Services, which runs about a third of the world’s online infrastructure, suffered a major failure that spread worldwide.
Apps from Snapchat to Roblox stalled, and millions of people met blank screens and timeouts. What began as a blip in AWS’s oldest region turned into hours of disruption, exposing how much it depends on a few central systems. As engineers rushed fixes and executives issued apologies, the message was blunt: in cloud computing, one provider’s glitch can stall the web.
The trouble began around 3 a.m. ET on 20 October in US-EAST-1, the Ashburn, Virginia region that has been AWS’s main workhorse since 2006. It hosts a vast number of services, from video streaming to finance. AWS’s status page cited a cascade of network faults linked to DNS, the plumbing that turns web addresses into IP numbers. When DNS fails, services cannot find each other, calls time out, and the stack grinds to a halt.
Specialists described a perfect storm. Error rates spiked in core database layers, and early-morning traffic pushed systems harder. “DNS issues are the silent killers of cloud ops,” said Ken Birman, a distributed systems professor at Cornell University. “They are basic, yet fragile.
AWS Outage Hit Globally
A mistake or overload in one Availability Zone can spread rapidly across dependent services.” The AWS Outage was the third big US-East-1 incident in five years, recalling the 2021 outage that disrupted airlines and the 2017 event that hit Capital One logins.
A full postmortem will take time, but early notes pointed to problems with load balancer health, throttled EC2 launches, and failover paths that could not cope. Traffic rerouting stalled, queues piled up, and many apps saw partial recoveries that then slipped back.
Amazon’s handling mixed rapid fixes with hard looks at architecture, not individuals. By 6 a.m. ET, teams had isolated the cause and started mitigation. They added capacity in nearby regions, restricted new virtual machine launches to steady the platform, and drained stuck message queues.
“We applied targeted fixes to network connectivity and are monitoring for full propagation,” said the AWS Health Dashboard around midday. Then came a second wave, hitting Lambda functions and S3 storage, which pushed full recovery out to about 6 p.m. ET.
By evening, the dashboard showed green across services. Amazon also waived some data transfer fees and offered credits, nodding to SLAs that promise 99.99% uptime or compensation.
The impact on people and businesses was vast. Downdetector counted more than 13 million reports by mid-afternoon, with U.S. spikes above 50,000 per hour. That undercounts enterprise pain.
Canada logged 351,000 reports, and apps like Wealthsimple stalled. In Europe, UK banks, including the Bank of Scotland, posted apologies as payments failed, and Premier League officials abandoned AI offside tools mid-game. CNET’s tally listed over 2,000 services affected worldwide.
Finance Apps Suffered
The list of casualties showed how deep the reliance goes. Snapchat’s 400 million daily users lost streaks and stories for hours. Roblox and Fortnite kicked players mid-session, cutting into playing time and sales.
Finance apps suffered, with Venmo transfers failing, Robinhood orders bugging out during wild markets, and Coinbase customers watching Ethereum slide without the option to trade. Amazon’s own products faltered too, with Ring missing doorbell presses, Alexa going quiet, and Kindle downloads pausing.
Delta and United reported app issues that delayed check-ins. Canvas locked students out of classwork. Perplexity said AWS caused its search outages. “The root cause is an AWS issue. We’re working on resolving it,” posted CEO Aravind Srinivas.
Zoom, Duolingo, Canva and Wordle saw problems. Signal’s downtime nudged some privacy-minded users to “AWS-free” options like Matrix. Streaming apps have been buffering for ages. Lunch-hour ordering broke for services like Lyft and Starbucks.
Analysts at Ookla put the economic hit from the AWS Outage in the hundreds of millions, with lost trades, abandoned baskets, and hours of lost productivity. “This is not just downtime; it is a sovereignty issue,” said Jake Moore, cybersecurity advisor at ESET. “We rely on three hyperscalers, AWS, Azure, and Google Cloud, and a wobble in one shakes everyone.”
For startups and engineering teams, the takeaway was harsh. Multi-cloud is not a buzzword; it is a safety play. Tools such as Route 53 DNS failover are available, yet many firms skip them to save cost or ship faster.
“If you cut corners on resilience, you are the one in the spotlight,” Birman said. Amazon highlighted its “well-established processes” in a statement, but critics see overconfidence. US-EAST-1 remains a chokepoint. Wider spread across regions and providers would soften the blow, yet inertia and vendor lock-in keep many put.
Regulators are likely to take notice after the AWS Outage. The EU’s digital sovereignty push will gain pace, and U.S. lawmakers may renew cloud antitrust interest. For now, services are back, but the question lingers in the background. How long until the next break? Outages are no longer rare. This AWS episode is not a footnote; it is a flaw in the system built around a few giant hubs.
