AWS DNS failure disrupted major apps despite services running internally
CrowdStrike and BGP outages failed at different infrastructure layers
Each outage exposed internet fragility through unique dependency breakdowns
At 3:12 AM Pacific Time (or 3:42 PM Indian Standard Time) on October 20, the internet’s invisible access lanes stopped working at the world’s largest online webhost situated in the USA. Amazon Web Services (AWS), the planet’s largest cloud infrastructure provider, suffered a major disruption in its US-EAST-1 region – the Northern Virginia zone that functions as the digital nerve centre for hundreds of websites and apps, from Reddit to Roblox to Snapchat. The culprit, according to Amazon’s own swift investigation and response, was identified as a DNS resolution failure.
SurveyWhenever it comes to explaining DNS (Domain Name System), it’s often compared to a phone book as an example of how it works. So essentially, the phone book of the internet went missing, and suddenly no one could remember how to call anyone anymore.
DNS failure caused AWS outage
For a slightly more technical read, here’s what actually happened. In the early hours of the morning (in the US, and afternoon here in India), AWS began observing elevated error rates across a bunch of services. EC2 (compute), Lambda (serverless functions), SQS (message queuing), and others. At the root was a failure in internal DNS – the service that translates human-friendly names like “dynamodb.us-east-1.amazonaws.com” into machine-readable IP addresses.
Think of it like losing your internal phone book. The servers are up, the databases are waiting, the applications eager to respond. But they’ve all forgotten each other’s numbers. Chaos ensues, but naturally.
And because this wasn’t just a glitch in an obscure tool but a failure in the core connected architecture of AWS’ internal universe, everything that relied on it began to stall, timeout, or simply error out.
Also read: Amazon AWS outage: Canva, Snapchat and major platforms are down, company responds
While AWS identified and addressed the DNS failure relatively quickly (around 3:30 AM PDT or 4:00 PM IST on October 20), the damage was done. Services had failed mid-request. Queues began to backlog. Lambda functions which are designed to respond in milliseconds were now hours behind in their resolution log. Instance launches failed. Logs were incomplete. And the impact was felt far beyond AWS’s console dashboards.
Snapchat users couldn’t send messages. Roblox players found themselves booted from servers or frozen on load screens. Reddit saw degraded performance and connection issues. These weren’t minor glitches as the internet’s favourite toys and platforms went dark.
How’s Oct 2025 AWS outage different from CrowdStrike in 2024 or FB BGP issues in 2021?
Of the past great internet failures, three events stand out for their impact and instructive contrast. The BGP-related outages that severed chunks of the internet from reality in 2021, the CrowdStrike-triggered Windows meltdown of 2024, and the recent AWS DNS failure that disrupted dozens of top-tier web services on October 20, 2025. Each broke the digital world in its own way – but they failed at fundamentally different layers of the internet’s architecture.
Short for Border Gateway Protocol, BGP is the map by which data knows how to travel between networks – the routing mechanism of the global internet. When Facebook accidentally withdrew its own BGP routes in 2021, the effect was profound. Its DNS servers became unreachable, and the rest of the internet quite literally forgot how to find Facebook. Similar routing accidents at Cloudflare and other backbone providers have had wide-reaching consequences in the past, not because the sites themselves were broken, but because the world had no directions on how to get to them.
Fast forward to July 2024, when cybersecurity firm CrowdStrike pushed out a faulty update to its Falcon agent, a widely-used endpoint protection tool. That software glitch bricked millions of Windows machines across the globe, plunging banks, airports, and government offices into chaos. It wasn’t that the internet was unreachable, but rather that people’s machines had been rendered useless in trying to access it.
Also read: CrowdStrike BSOD error: Risking future of AI in cybersecurity?
Compare that to what happened in October 2025 with AWS, this time the failure was nestled deep inside the cloud – specifically, inside AWS’s own internal DNS system in the US-EAST-1 region. Within AWS, it also helps services discover each other. So when that system faltered, services like EC2, DynamoDB, and Lambda couldn’t find or talk to each other, despite technically being online.
Here’s one way to describe the difference between these three types of internet outages. BGP outages broke the roads of the internet. The CrowdStrike event stalled the vehicles. The AWS outage was like an office building where all the employees showed up for work, but the company directory was wiped – nobody knew where anyone else sat, or how to reach them. And while the surface experience for users – Snapchat or Reddit down – may sound similar, the lessons from each of these failures are distinct, pointing to the complex, fragile interdependencies that power our always-online world.
Follow Us
Jayesh Shinde
Executive Editor at Digit. Technology journalist since Jan 2008, with stints at Indiatimes.com and PCWorld.in. Enthusiastic dad, reluctant traveler, weekend gamer, LOTR nerd, pseudo bon vivant. View Full Profile
