Early on the morning of October 20, the cloud quietly sneezed, and half the internet caught a cold.
At around 03:11 EST (08:11 BST), AWS’s US-East-1 region began experiencing increased error rates and latencies, according to the AWS service health page.
The initial internal DNS failures caused a cascade of disruptions across services from enterprise SaaS to messaging, banking, education, and even smart home systems. While Amazon applied a fix, API problems resulted in a wave of secondary outages.
It served as a reminder that infrastructure failures still translate to customer frustration — and, more importantly, a sudden loss of trust.
When everyday services like banking apps, smart devices, and communication tools stop working, users don’t think about DNS errors or data center regions; they just know something they rely on has failed them.
Users of social and messaging platforms like Snapchat, Signal and Reddit quickly flooded outage trackers with reports of frozen feeds and error messages.
The disruption soon rippled into more critical sectors. UK customers of Lloyds and Halifax couldn’t log into their accounts or complete transactions, while users of Amazon’s own Ring doorbell and camera were left staring at unresponsive devices.
Even education and enterprise tools weren’t spared. Universities like Rutgers reported degraded performance across learning portals, Zoom and Adobe platforms.
Together, the incidents underscored just how deeply everyday digital life depends on cloud reliability, from morning meetings to mobile payments and multiplayer games.
By mid-morning in the EST time zone, Amazon’s website reported 36 of its services were back up and running, including Amazon Connect, but another 72 were still interrupted.
“We have applied multiple mitigations across multiple Availability Zones (AZs) in US-EAST-1 and are still experiencing elevated errors for new EC2 instance launches,” the company stated.
The update signaled that even as services began to recover, new workloads and restarts remained unstable, a clear sign of how deeply the issue had rippled through the system.
“Similar failures have been common causes for major outages in the past, and usually stem from incorrect, updated configurations, or due to poor monitoring of expiration timelines for configurations, certificates, etc.,” said Aras Nazarovas, Senior Security Researcher at Cybernews.
“From initial reporting there are no indications of any security breach, however failing to keep information or resources available for clients can be classified as a cyber incident, even if there was no malicious outsider or malicious intent.”
Nazarovas noted that, while not malicious in nature, such configuration-related missteps can affect the customer experience for thousands of businesses.
Similar outages occur almost every year, and they can be a reminder of how extensive software supply chains have become, showing how a simple issue on a handful of Amazon Data Centers caused thousands of issues to their clients… in worst case scenarios such an outage could have had serious consequences in critical infrastructure sectors.
The outage also highlights the importance of disaster preparedness in coordinating recovery and maintaining communication with customers, Nazarovas said.
“In the event of such disruptions, users should immediately seek alternative solutions for communication (different app, phone calls, SMS, radio) to be able to coordinate next steps towards recovering from such a disruption. It is a good practice to have a ‘Disaster Recovery Plan’ where alternative communication channels and other critical steps have been planned in advance.”
The mood on social media was equal parts frustration and humor. Users churned out memes such as “And this is why everyone using the same damn Amazon web services is a great idea” and a graphic showing all the sites that depend on AWS.
But the humor reflected a broader commentary on how heavily enterprises depend on a handful of cloud providers.
The incident offers several clear lessons. Downtime translates directly into customer frustration. When users cannot access the services they expect, the perception is immediate.
And trust is fragile. If a bank customer trying to log in at 09:00 finds “service unavailable”, trust erodes. Customers of highly sensitive services like banking expect 100% uptime. Even if it is not caused by the bank’s own systems, an outage still hits its bank reputation.
The fact that an infrastructure outage (DNS/DynamoDB endpoint in US-EAST-1) affected downstream services brings home that indirect dependencies matter when it comes to customer experience and their perception of a service provider.
Communication is key for the customer. Some services were quick to post status updates, while others were not. Having a clear communication plan is now a core part of providing a highly-regarded experience.
Enterprises should know where their dependencies are and have updated contingency plans. They should also expect issues such as failed payment authorizations and spikes in customer support contact for days or even weeks after such an outage.
