A technical failure at Amazon Web Services (AWS) keeps the service inoperative this Monday, October 20, 2025, affecting hundreds of global websites and applications. The issue, starting around 3 AM (Brasília time), stems from a disruption in the Domain Name System (DNS) in the US-East-1 region, the company’s main hub in the United States. Companies like Snapchat, Roblox, and United Airlines report impacts, with users facing unavailability of essential services.
The outage highlights the dependency of digital infrastructure on a few cloud providers, where a single error can cascade failures to thousands of clients. AWS technicians identified the root cause, but full recovery requires processing a backlog of queued requests. Millions of users, especially during peak morning hours, face difficulties accessing bank accounts, flight bookings, and online games.
Several sectors report immediate losses due to the prolonged downtime.
- Payment platforms like Venmo and Coinbase block transactions.
- Learning apps like Duolingo halt lessons and syncs.
- Streaming services, including Apple Music, fail to play audio.
Cybersecurity experts monitor the incident, ruling out signs of a cyberattack so far.
Origin of the DNS disruption
The failure began when the DNS, responsible for translating web addresses into IP numbers, lost communication with databases like AWS’s DynamoDB. This region handles high global traffic, amplifying the issue’s reach.
Companies reliant on this infrastructure face delays in accessing stored data. AWS recommended that clients clear temporary caches to speed up restoration.
Affected services in detail
Snapchat reports user complaints about inability to load photos or messages in real time.
Roblox and Fortnite interrupt gaming sessions for millions, forcing pauses in online matches.
United Airlines reports app and website failures, complicating check-ins and flight updates for passengers in the US.
Coinbase confirms funds remain secure, but platform access is blocked during the outage.
Delta Airlines notes similar issues, with booking systems operating at limited capacity.
Ongoing mitigation efforts
AWS implemented parallel fixes to restore the affected regional gateway.
By around 10 AM (Brasília time), the company observed signs of recovery in some services, with most requests processed successfully.
Engineers are addressing a backlog of pending commands, which may cause residual slowdowns in virtual applications like EC2.
The company plans a detailed report in the coming days, outlining the failed steps and preventive measures adopted.
History of similar outages
Previous AWS incidents occurred in 2021, lasting hours and impacting e-commerce.
In 2024, a CrowdStrike failure caused $5 billion in global losses, highlighting vulnerabilities in software updates.
Defective updates or incompatibilities with third-party tools are common causes of such disruptions.
The relative frequency of these events reflects challenges in redundancies, according to IT analysts.
The concentrated reliance on a few cloud regions amplifies risks, with local failures quickly spreading to interconnected networks. AWS holds about a third of the global cloud computing infrastructure, making its outages global-scale events. In Brazil, services like Mercado Pago and Mercado Livre reported mild instabilities, resolved by around 7:30 AM, per real-time monitoring. This concentration underscores the need for diversification, though investments in regional backups have reduced average outage durations from days to hours in recent years. Full recovery depends on rigorous testing in future updates, ensuring components like DNS operate without single-point dependencies.
Recommendations for users
Maintain alternative connections for essential services, such as offline apps or local backups.
Monitor AWS’s official status pages for real-time updates.
Companies should review contingency plans, testing migrations to multiple cloud providers.