AWS CloudFront incident knocks numerous global websites offline for hours

On Thursday (June 16), a routing incident deep within the Amazon Web Services (AWS) system abruptly disrupted the operations of thousands of websites worldwide. For over three hours, users attempting to access these platforms received only system error messages, triggering a minor wave of internet chaos before Amazon timely contained and resolved the situation.

www.ipsip.vn

Cybersecurity in Vietnam Q1/2026: Data leaks spike, APT and DDoS escalate

Cybersecurity in Vietnam Q1/2026 recorded 165 data breaches, over 473 million leaked records, alongside surging DDoS and APT attacks. Complete overview of trends and enterprise recommendations.

What happened to the websites?

According to reports from the AWS service health dashboard, the incident began early Thursday morning (UTC-7). Numerous customers using the CloudFront service (a system that supports web content delivery and acceleration) combined with the VPC Origins connectivity feature simultaneously reported that their websites were returning 5xx errors—a technical term indicating that the server encountered an issue and could not respond to user requests.

To make it simple, many large enterprises often choose to place their sensitive applications and data inside a "digital safe zone" called a Virtual Private Cloud (VPC) to hide all servers and IP addresses from the public Internet to prevent attacks. In this setup, the VPC Origins feature acts as a secure bridge, allowing the CloudFront service to enter this private zone to fetch data and display it to users.

However, during this incident, the aforementioned bridge broke down. An Amazon representative stated that their internal connection management system had hit a physical limit threshold. As a result, the subsystem responsible for distributing routing configuration data to network processors failed, preventing CloudFront from communicating with customers' internal secure servers, directly knocking down front-end websites.

Global impact spreads widely

Although confidentiality clauses do not allow Amazon to publish a detailed list of affected customers, the sweep of this incident was significant. Data from the research platform TheirStack shows that more than 6,000 companies in North America, 2,100 enterprises in Europe, and over 1,300 entities in Asia are operating on Amazon's private cloud platform.

Among the victims directly affected by this disruption, tech publication The Register pointed out the artificial intelligence development platform HuggingFace and the website of the UK National Lottery. In addition, tech services such as Tailscale (affecting the admin console and package repository) and Ubiquiti (connectivity errors across multiple cloud services) also officially confirmed encountering issues. Data from the outage tracking website Downdetector also recorded hundreds of error reports from users during the same timeframe.

The incident also drove a wave of anger from tech administrators onto the social media platform X (formerly Twitter). Many fiercely criticized AWS support for leaving the system "paralyzed" for two hours before posting warning information on the status page, leaving customers' operations engineers completely in the dark.

Data from the outage tracking website Downdetector also recorded hundreds of error reports from users during the same timeframe.

Potential risks of "putting all eggs in one basket"

This incident once again sounds the alarm about the structure of the modern Internet. Mr. Mayur Upadhyaya, CEO at APIContext, commented that it is entirely normal and unavoidable in the tech industry for a cloud infrastructure provider to experience a technical outage. However, the concern lies in the current shift in trends.

"The more concerning trend is that we're increasingly consolidating around a small number of providers because they're the most convenient and economically attractive choice... that consolidation changes the nature of operational risk. A fault that might once have affected a handful of organizations can now impact thousands simultaneously because so many businesses depend on the same infrastructure.", Mr. Upadhyaya analyzed.

The expert also emphasized that the most important question right now is not when the next incident will occur, but whether businesses' core transactions have backup mechanisms to sustain operations when that worst-case scenario repeats.

Costly lessons from the past

This is not the first time cloud computing giant AWS has sent shockwaves through the tech world. Looking back at recent operational history, in May, a cooling system failure at a data center in the Northern Virginia region froze the trading activities of the Coinbase exchange and CME Group.

More seriously, in October of last year, a Domain Name System (DNS) error in AWS's US-EAST-1 region wiped a series of popular applications off the Internet for nearly an entire day. During that time, messaging, entertainment, and social media applications such as Signal, Snapchat, Reddit, Fortnite, and Ring, along with Apple TV and Apple Music services, all stopped working. Even smart home devices like the Eight Sleep high-tech bed or the Amazon Alexa virtual assistant lost connection and malfunctioned.

This is not the first time cloud computing giant AWS has sent shockwaves through the tech world.

Following efforts to deploy phased response measures, Amazon engineers confirmed the system was fully recovered at 5:21 AM (UTC-7). Amazon also announced that customers who had applied temporary workarounds (switching to a different type of configuration source during the incident) could now safely switch back to their original configuration. Although the incident has been completely resolved, it remains a practical reminder for businesses about the importance of building flexible backup scenarios in the digital environment.

---

Reference: The Cybernews