Cloudflare incident on August 21, 2025

Cloudflare describes a regional outage on 2025-08-21 where a surge of traffic from a single AWS us-east-1 customer saturated peering links, causing congestion, elevated latency, packet loss, and failures to origins for customers using AWS us-east-1; the event was operational in nature (not an attack or BGP hijack). The issue was mitigated through rate limiting, traffic engineering, and coordination with AWS, with full normalization by 20:18 UTC, and Cloudflare plans DCI capacity upgrades, per-customer traffic deprioritization, and a new traffic management system to prevent similar incidents.