Largest Outages of 2025: A Downdetector Analysis | Ookla®

In his latest Ookla analysis, Perry Haghighi details how 2025 became a year of “Logic Layer” failures rather than hardware crashes. For Security and IT professionals, these stats illustrate the terrifying reach of centralized dependencies: The AWS mega-outage on Oct 20 (US-EAST-1) was the year’s largest, generating 17 million global reports. Traced to automated DNS management for DynamoDB, the 15-hour event crippled “isolated” apps like Snapchat (944k reports) and Netflix, proving that control plane logic is now our biggest single point of failure. Gaming and streaming weren’t far behind. The Feb 7 PlayStation Network blackout lasted 24 hours with 3 million reports, while YouTube’s Oct 15 streaming glitch saw 1.5 million. Regional resilience also fluctuated, with X (Twitter) peaking at 645k reports in APAC and WhatsApp hitting 621k in LATAM. The takeaway for architects is clear: automation and centralization have created a “Thundering Herd” risk. When providers like Cloudflare or AWS hiccup, the MTTR is no longer in your team’s hands, but in the logic of a third-party global control plane.