AWS Outage 2025

The October 23, 2025, report by The Register details a 15-hour AWS outage triggered by a "race condition" in DynamoDB’s automated DNS management.

Lets summarize the key points:

- The Cause: A conflict between two internal systems (the "Planner" and "Enactor") caused an automated cleanup script to accidentally delete the IP addresses for DynamoDB in the US-EAST-1 region.
 

- The Impact: Because DynamoDB is a core dependency, it crippled other services like EC2, IAM, and Lambda. This sidelined everything from global banking to smart home devices (Ring, Peloton).

- The "Traffic Jam": Recovery was delayed by "congestive collapse," where millions of devices trying to reconnect at once overwhelmed the system.
 

- The Fix: Amazon has disabled that specific DNS automation and is adding "guardrails" to prevent automated scripts from making such destructive changes in the future.

 

[1] https://www.theregister.com/2025/10/23/amazon_outage_postmortem/?td=rt-3a