AWS. Google Cloud. Azure. CrowdStrike. All hit hard by dumb bugs with big blast radii - race conditions, nulls, misfired configs. Small cracks. Massive fallout.
AWS's DNS automation knocked out its DynamoDB endpoint, dragging 113 services down with it. Google Cloud’s global APIs fell over from a stray null pointer deep in quota policy code. CrowdStrike pushed a config that boot-looped 8.5M Windows boxes in kernel mode. Azure let a bad config sneak past checks and spread it across its global AFD fleet.
System shift: These weren’t freak accidents. They’re signs of strain in how we scale - centralized control planes, eager global replication, recovery playbooks that break when too much breaks at once.
