When the Internet Stops: the Amazon Web Services Outage and the Hidden Fragility of the Modern Web

On 20 October 2025, shortly after 1 pm US Eastern Time, a series of anomalies in the US-EAST-1 region of Amazon Web Services (AWS) triggered one of the most serious service outages in recent years. Within a few hours, millions of users around the world found themselves unable to access websites, cloud platforms, payment services and smart devices, including Amazon’s own.

According to the technical report published by AWS, the outage originated from an error in the Domain Name System (DNS), the mechanism that translates domain names (e.g. amazon.com) into IP addresses that servers can understand. When DNS does not respond correctly, applications and services can no longer “find” each other on the network.

The problem was made worse by a malfunction in the load balancers, the components that spread traffic across servers to prevent overload. In this case, one of the load-balancing monitoring systems started wrongly flagging nodes as “unreachable”, effectively cutting them off from the network.

The result: part of AWS’s data centres stopped communicating properly with the others, generating a cascade of internal errors and unresolved requests.

What happened?

AWS explained that it all started with an internal update to a network service used by DynamoDB, the NoSQL database that acts as the backbone for thousands of apps and websites.

The update, introduced to improve the system’s resilience, instead created an unexpected condition that saturated internal DNS resources, slowing down responses to connection requests.

When the internal DNS could not respond, the load balancers, designed to redistribute traffic automatically, began to “see” servers as unavailable and shifted huge volumes of requests onto other nodes, which in turn quickly became overloaded.

This created a loop of progressive degradation, in which every attempt at self-regulation made things worse.

AWS took about six hours to restore full service, temporarily switching off some automated systems and manually restoring DNS and routing settings.

A global and visible impact

The shockwave was immediate:

• Fortnite, Snapchat, Alexa, Ring, Zoom and Disney+ went offline or suffered significant delays.
• Several banking services and payment apps (including international gateways) saw transactions blocked.
• Even smart home platforms, such as Ring cameras and doorbells, were inaccessible for hours.

In Europe, and to a lesser extent in Italy, the impact was indirect but tangible: slowdowns, login failures and synchronisation problems on services that use US-based AWS infrastructure as their back end.

Infrastructure concentrated in the hands of a few

Today around 70% of the European cloud market is controlled by Amazon, Google and Microsoft. This concentration ensures efficiency and performance, but it also represents a potentially catastrophic single point of failure.

The ideal of a “decentralised”, resilient and distributed Internet is now a distant one: most of the digital services we use every day run over the same backbones, the same data centres and even the same DNS systems.

In this scenario, a configuration error or a simple bug can turn into a worldwide blackout.

The lesson of the outage: resilience and diversification

The AWS outage was not just a technical incident but a wake-up call.

Businesses that rely on the cloud will need to invest in multi-cloud strategies, spreading workloads and data across several providers, and in disaster recovery plans able to guarantee business continuity even in the event of a total failure.

A global network… but a fragile one

We live in a world where our daily lives , from paying for a coffee to switching on a smart light bulb , depend on invisible, interconnected systems.

The AWS outage in October was a mirror of this fragility: a small technical error, at a single point in the network, showed how thin the thread holding up the whole digital infrastructure really is.

The question is not whether it will happen again, but whether we will be ready to face it.