Amazon’s AWS Disruption Creates Outages for Hundreds of Websites for Hours

A disruption involving Amazon Web Services, the cloud service provider that supports much of the internet, took many websites and apps offline for over two hours on Monday, in the latest outage that showed the fragility of the global technology infrastructure.

The disruption, which affected websites and apps for some major banks, gaming sites and entertainment services, started shortly after 3 a.m. Eastern. Amazon said in an update at 5:27 a.m. that most websites and apps relying on its services were working normally again, and that it continued “to work through a backlog of queued requests.”

Major services were affected, including WhatsApp, the British government’s website and government tax services, the payment app Venmo, the cryptocurrency platform Coinbase and games at The New York Times. Dozens of other companies and retailers — including Amazon, Venmo, Hulu, Snapchat, Ring doorbells and McDonald’s — also experienced service interruptions.

It was not immediately clear what led to the outage, and there were no indications that it had been caused by a cyberattack.

Experts said that the disruption showed again how the internet’s reliance on a few major technology providers — including Amazon, Microsoft and Google — can mean disruptions for millions of users when one service breaks down. Last year, a much wider, daylong internet outage was caused by a faulty update sent out by a little known cybersecurity company called CrowdStrike.

Amazon Web Services counts thousands of clients who rely on it for complex, demanding, data-intensive operations, including streaming video, running web applications and storing huge amounts of digital information. Amazon’s cloud-computing division has set up infrastructure all around the world, allowing companies to make their products accessible to customers across the globe. By using Amazon’s service instead of building their own, clients can scale up or down without having to invest heavily in costly hardware.

In its initial statement on the outage, Amazon’s said early Monday that 28 of its services, including those in the “US-EAST-1” region, were having issues and that its engineers had been working on limiting the effects and identifying the cause.

Harry Halpin, the chief executive of NymVPN, a virtual private network service, said that the issue could have been triggered by a technical fault in one of Amazon’s main data centers. But he added that cloud platforms’ operations are inherently opaque, making it impossible to know the cause unless Amazon disclosed it, including if the cause was a cyberattack.

Mr. Halpin, whose company provides VPN services to soldiers in Ukraine, said he woke up to several emails from soldiers on the front lines asking what had caused the disruption. The problem extends beyond Ukraine and applies to other Western governments, many of which rely on such cloud services, he said.

“If your entire nation’s infrastructure relies on a few providers, all in the United States, and anything can go down at any moment, either for malicious reasons or just technical errors, that’s an exceedingly dangerous situation,” he said.

“Everyone takes it for normal,” said Dr. Halpin, a former research scientist at the Massachusetts Institute of Technology, referring to the internet’s dependence on a handful of providers. “But it’s not normal.”

Some media advocates said that the outage, which caused disruptions to secure communications apps such as Signal and other digital tools, showed how the internet’s reliance on a few major technology companies posed a risk to free speech.

“When a single provider goes dark, critical services go offline with it,” Corinne Cath-Speth, head of digital for Article 19, a free speech advocacy group, said in a statement. She added that there was an urgent need for diversification in cloud computing.

“The infrastructure underpinning democratic discourse, independent journalism and secure communications cannot be dependent on a handful of companies,” she said.

Still, Amazon’s share price barely moved in premarket trading, suggesting that investors were not too bothered about the outage. In the first half of the year, Amazon Web Services accounted for nearly 20 percent of Amazon’s sales, but about 60 percent of its operating profit.

Melissa Eddy in Berlin and Andrés R. Martínez in Seoul contributed reporting.