Amazon Web Services: AWS back to normal operation

Yesterday, Monday, a kind of internet catastrophe (greatest plausible disaster) occurred: an error in Amazon's server infrastructure has numerous dependent internet services paralyzed. This affected not only Amazon's own offerings like Prime, Music, and similar, but also providers who booked capacity in Amazon's cloud. Around midnight, Amazon reported a return to normal.

In the entry on Amazon's cloud status website, the company now explains the causes of the outage and their solution. The summary in the entry is dated 0:53 AM Central European Summer Time on October 21, 2025. According to the entry, between 8:49 AM local time and 11:24 AM on October 20, 2025, there were “increased error rates and latencies in AWS Services in the US-EAST-1 region.” Additionally, services and functions dependent on US-EAST-1 endpoints, such as IAM and DynamoDB Global Tables, experienced problems during this period.

“Around 9:26 AM, we were able to identify the trigger of the incident as DNS resolution issues in the regional DynamoDB service endpoints. After resolving the DynamoDB DNS issues around 11:24 AM, services began to recover, but subsequent impairments of internal EC2 subsystems responsible for launching EC2 instances occurred due to their dependency on DynamoDB,” Amazon further explains.

Problems during troubleshooting

As Amazon technicians worked through the EC2 instance startup issues, network load balancer health checks – i.e., network monitors – were also impaired, causing further connection disruptions in several services such as Lambda, DynamoDB, and CloudWatch. The load balancer health checks were brought back under control around 6:38 PM.

“As part of the efforts to restore services, we temporarily throttled some operations, such as EC2 instance startup, processing of SQS queues via Lambda Event Source Mappings, and asynchronous Lambda calls. Subsequently, we reduced the throttling and worked in parallel on the network connectivity issues until the services fully recovered. At 0:01 AM on October 21st, all AWS services returned to normal operation,” Amazon's technical team further states.

Some services like AWS Config, Redshift, and Connect still have a backlog of messages that they will process in the coming hours. Amazon announces a more detailed report as a post-mortem of the event.

(dmk)