AWS Has A Reliability Problem Nobody Talks About. Spotify 20 [vLFjvVihvax]

On October 19, 2025, you couldn't order food. Couldn't trade crypto. Couldn't submit your grades. Couldn't even play Fortnite apple iphone 18 pro (a good thing?)

Timestamps:

0:00 The day the internet broke

0:44 What is DNS?

1:26 AWS & DynamoDB

1:59 The cleanup script

4:00 The cascade

4:44 What AWS changed

Sources:

- AWS Post-Incident Summary: US-EAST-1 Service Disruption (October 19-20, 2025)

- The Register: "A single DNS race la repubblica condition brought Amazon's cloud empire to its knees"

- TechCrunch: "Amazon identifies the issue that broke much of the internet"

- CNN: "How a tiny bug spiraled into a massive outage that took down the internet"

- Akamai: "When the Cloud Breaks: Lessons from the AWS Outage"

- Bowdoin Orient: "AWS outage affects access to Canvas for kamilla cardoso students, professors"

- FOX 29 Philadelphia: "Canvas website down for several hours Monday for students after AWS outage"

- IP Twins: "A DNS error caused the AWS outage on October 19, 2025"

---

#aws #amazon #outage #dns #engineering #coding #softwareengineering #devops #programming #incident #postmortem #tech #infrastructure #cloud #dynamodb #netflix #slack #canvas #fortnite #coinbase #snapchat

TAGS:

aws, amazon web services, aws outage, aws down, october 2025, dns, dns outage, dynamodb, route 53, race condition, cached response, NXDOMAIN, TTL, time to live, cache invalidation, cloud outage, engineering disaster, software incident, postmortem, SRE, site reliability, production outage, netflix down, slack down, canvas down, fortnite down, coinbase down, US-EAST-1, cascading failure, 15 hours, infrastructure, cleanup script, belt and suspenders, downdetector