このエピソードについて
Behind the story: A Google data centre in London had duplicate power supplies, emergency generators and battery protection. Yet on 20 October 2026, the generators failed to start because the systems controlling them misunderstood whether electricity was actually available.
The resulting outage affected services beyond the original failure.
For Michael, the incident recalled a Friday morning years earlier when an upstream supplier suddenly ceased trading, leaving every customer of his software business without service. By lunchtime, his team had restored the first customer using servers on another continent. The others followed before the end of the day.
Two different incidents reveal the same fundamental question: what makes an organisation genuinely resilient?
In this episode
- Why Google's emergency generators failed to start
- How a failure in one availability zone affected services elsewhere
- What happens when a supplier to your supplier disappears
- Why responsibility to customers continues even when an outage isn't your fault
- The difference between preserving information, restoring operations and retaining understanding
- Why a backup isn't necessarily an independent alternative
