Twelve Hours of Quiet
The Optus outage, a routine upgrade, and what happens when your safety mechanisms all trip at once.
Series
The Optus outage, a routine upgrade, and what happens when your safety mechanisms all trip at once.
UniSuper and Google Cloud: geographic redundancy doesn't protect you from a deletion that hits everything at once.
One week after the CrowdStrike outage, and it happens to be SysAdmin Day. Fitting.
CrowdStrike's root cause analysis shows how a mismatch between 20 and 21 took down millions of machines.
OpenAI's December outage: a monitoring service took down the platform it was meant to monitor.
Heathrow closed for most of a day after a fire at one of its three power supply points. Redundancy is about the switchover.
Spain and Portugal lost power in seconds. Every resilience plan assumes the grid is there.
Google Cloud's 12 June outage came down to an unchecked null, and code that had no feature flag.
The AWS us-east-1 outage: a race condition, a deleted DNS record, and a very long recovery tail.
Cloudflare's 18 November outage: a database permission change, a file that doubled in size, and a hard limit.