A hiccup at Amazon Web Services briefly broke half the internet
The widespread failure highlights just how centralized the internet-as-we-know-it has become. It also highlights AWS’s critical role
Photo by Matthias Balk/picture alliance via Getty Images
For a few hours early Monday morning, much of the internet seemed to short-circuit.
Suggested Reading
An outage originating with Amazon Web Services (AWS), the system that powers a truly breathtaking swath of online life, swept across hundreds of major sites and platforms, including Facebook, Amazon, Snapchat, Roblox, Reddit, Venmo, Slack, Zoom, Hulu, Roku, Coinbase, and even Wordle.
Related Content
The Starbucks app went down, too, so that customers trying to order their early-morning caffeine fix found the routine task at least briefly impossible. The McDonald’s app was similarly affected.
East coast epicenter
The trouble appears to have begun around 3 a.m. E.T. in Amazon’s “US-EAST-1” data region, a cluster of servers in Northern Virginia that handles traffic for much of the East Coast. A malfunction in the company’s DynamoDB system (essentially a cloud-based database engine) sent error rates surging across dozens of AWS products.
That single failure rippled outward fast. Apps crashed. Payment platforms froze. Delta and United check-in kiosks blinked out, too. Media organizations attempting to cover the crash, including the Wall Street Journal, found their own sites affected.
Downdetector, which monitors site disruptions, listed hundreds of affected companies. By 5:30 a.m., Amazon said it was "seeing significant signs of recovery." But for some, services remain sluggish. Reddit, as of this writing, is still struggling to come back online.
A cloud that casts a long shadow
The widespread failure highlights just how centralized the internet-as-we-know-it has become. It also highlights AWS’s critical role. AWS hosts millions of sites and provides the computing muscle for much of the global economy, from gaming to banking, streaming entertainment to healthcare. Restaurant and ride-hailing apps, too.
In other words, when AWS falters, the impact isn’t limited in scope. It becomes the equivalent of a major regional power outage.
It’s also not the first time this has happened. AWS has suffered similar incidents in recent years, each one revealing how the internet’s existence largely depends on just a handful of tech giants.
You might say that a cloud can cast a long shadow.