The Internet has just experienced a new global outage. Why does this keep happening?

new York
—

Amazon Web Services (AWS), the cloud computing platform that powers much of the Internet, went down for several hours on Monday, causing several major consequences. Unusable websites and apps.

From banking and social media to airline booking sites and online shopping, thousands of services were disrupted as millions of people around the world – many commuting to work on the US East Coast – were unable to order coffee on their mobile or access key apps.

The latest outage is a reminder of how fragile the backbone of the Internet can be, even if the disruption is brief, and how dependent the world has become on these online services.

Although AWS and its competitors are generally robust, the Internet is a complex network of overlapping services whose reliability depends on their weakest code. The root cause of Monday’s outage remains unknown, but a service that converts user-friendly web names into IP addresses was unable to communicate with the massive databases of thousands of businesses hosted by Amazon.

Past outages of this magnitude have been caused by a wide variety of errors, including faulty updates, accidental injection of bad code, or a modification of third-party software that doesn’t work well with a service. Rarely, Internet cable outages, cyberattacks, or direct denial of service attacks can bring down or overload servers that host key applications.

But the relative frequency of these events shows the lack of necessary layoffs and competitive services. Too often, some Internet experts say, companies put all their eggs in one basket of cloud services.

There are “no signs” that this is a cyberattack, according to Rob Jardin, chief digital officer of cybersecurity firm NymVPN, adding that it “looks like a technical fault affecting one of Amazon’s main data centers.”

“The Internet was originally designed to be decentralized and resilient, but today much of our online ecosystem is concentrated in a small number of cloud regions,” he said in a note. “When one of these regions experiences an outage, the impact is immediate and widespread.”

Jardin said that “these problems can arise when systems are overloaded or a key element of the network fails; and because many websites and applications rely on AWS, the impact spreads quickly.”

AWS doesn’t often experience major disruptions like this, the latest one occurring in 2021.

“It’s comparable to other major cloud providers and, in fact, it’s amazing that they are able to operate at such a scale without more frequent interruptions,” said Mike Chapple, a cybersecurity expert and professor of computer science at the University of Notre Dame’s Mendoza College of Business.

“The reason these events get a lot more attention is their impact,” he told CNN. “If just one company has a problem in its data center, it leads to problems for that company’s products and services. »

In 2024, the the biggest computer outage ever seen destroyed large parts of the Internet when a devastating CrowdStrike software glitch crashed computers, led to flight cancellations and disrupted hospitals around the world, creating $5 billion in direct business losses. A bug in CrowdStrike’s cloud-based testing system led to a problematic update on computers around the world.

Also last year, AT&T’s network has been down several timesincluding an 11-hour crisis that left many gig workers unable to do their jobs.

AWS is a cloud computing provider that hosts many of the world’s most widely used online services. In its early days, Amazon needed excess server capacity to ensure it had enough computing power to handle the massive amounts of traffic coming to its site during the holiday season. Amazon realized that during the rest of the year it could use these servers to serve the online needs of other businesses, and thus AWS was born.

Among AWS’s many offerings is DynamoDB, a database that hosts information for businesses, including customer data. Amazon said on Monday that its customers could not access data stored in DynamoDB because the domain name system (DNS) – a kind of telephone directory for the Internet – had encountered a problem.

DNS is like an Internet location engine, converting friendly web addresses like amazon.com into IP addresses – a series of numbers that other websites and applications can understand.

“Amazon stored the data securely, but no one else could find it for several hours, leaving the apps temporarily separated from their data,” Chapple said. “It’s as if large parts of the Internet are suffering from temporary amnesia.”

It’s unclear what exactly caused the DNS outage, but it only lasted a few hours. As of 6:35 a.m. ET, Amazon had resolved the DNS issue and recommended businesses clear their cache (temporary storage files) to speed up the restoration of their services.

Amazon said the outage continued to affect other AWS services, including EC2 – a kind of virtual server that many companies use to build their online applications.

The company will likely conduct a post-mortem and explain what went wrong with its DNS system in the coming days.