Every major cloud outage, broken down and scored for SLA-credit eligibility - what broke, how long it ran, and what your provider owed you.
A DNS resolution fault for the regional DynamoDB endpoint in us-east-1 broke DynamoDB itself and the many AWS control planes that depend on it, cascading to EC2 launches, Lambda, IAM, and hundreds of third-party apps. DynamoDB is covered by a 99.999% multi-Region and 99.99% single-Region SLA, so even a few hours of regional error rates cleared the top credit tier for single-Region tables.
8 min read
On June 12, 2025, an invalid automated quota policy update propagated globally and caused Google Cloud API requests to fail with 503 errors across dozens of services and regions, cascading to Cloudflare, Spotify, and others. Because most Google Cloud services publish 99.95% or 99.99% monthly SLAs, even a few hours of global API errors cleared the first credit tier for affected customers.
8 min read
A backend storage and networking failure in the Azure Central US region on July 18, 2024, took virtual machines and dependent services offline for several hours. Single-instance and Availability Set VMs are covered by 99.9% and 99.95% monthly SLAs, so a multi-hour regional failure cleared the first Azure credit tier for affected customers, separate from the unrelated CrowdStrike outage a day later.
8 min read
On November 25, 2020, a routine capacity addition to Amazon Kinesis in us-east-1 pushed its front-end fleet past an operating-system thread limit, and the fleet fell over. Kinesis is covered by a 99.9% monthly SLA, and the many services that depend on it (CloudWatch, Cognito, Lambda event sources, and more) degraded too, so the multi-hour breach cleared the first credit tier for a wide set of customers.
8 min read