The zone that went down in London was a power panel
On 20 October 2026, at 06:08 UTC, a "power event" inside a single zone — europe-west2-b, one of the three in Google Cloud's London region — took banks, parcel services and listing portals offline for nearly eight hours. No region went down: the power went down inside one zone. And "we are spread across two zones" is not an answer, because a zone name is a distribution label, not a place you can audit: AWS documents that us-east-1a is not the same physical location in two different accounts. We explain why the failure domain that costs the most is the one nobody draws, why Ceph's default hierarchy has had a bucket type called pdu for years, when NOT to raise your failure domain, and the one-row-per-replica map we fill in standing next to the rack.
Microsoft 365
ES 
CAT 
EN