Technology · NEWS EXPLAINER
Google Cloud restores London region after power event disrupts UK banks
The europe-west2 disruption began at 06:08 UTC and hit banking, delivery and public-sector services. Google said full service was restored later Tuesday, while promising a detailed incident analysis.
Google Cloud said full service was restored in its London-based europe-west2 region on October 20 after a power event caused hours of elevated errors and intermittent network failures. The disruption spread beyond cloud dashboards to banking, delivery and public-sector services.
What happened
Google Cloud reported a service disruption in its London region, europe-west2. AP placed the start just after 06:08 UTC, while Google described the trigger as a power event. The company said the incident produced elevated errors and intermittent network failures across multiple services; it has not yet published the promised full technical analysis.
Why users noticed it outside Google Cloud
Reuters reported disruptions at Revolut, Lloyds, HSBC, Rightmove, DPD, NHS England and Royal Mail. Revolut said card payments and ATM withdrawals were affected but customer money and data remained safe. Royal Mail said letters and parcels were not affected, showing that a reported outage at an organization did not necessarily disable every operation.
The recovery signal
Google said engineering teams restored full service availability later Tuesday. Lloyds said its app was operational again and expected the rest of its services to normalize. AP said the main disruption lasted roughly six hours, but customers should rely on their own provider status pages because application recovery can lag behind the underlying cloud region.
What remains unknown
The phrase “power event” identifies the immediate infrastructure category, not a complete root cause. Google said it will publish an incident analysis after its investigation. There is not yet a public explanation of the failed component, the behavior of backup systems, the exact customer count or whether every dependent application recovered at the same time.
Practical resilience lesson
A multi-zone or multi-region design only helps when data, identity, networking and failover procedures are tested together. Organizations should verify which dependencies still point to one region, document a customer-communication path that does not share the same failure domain, and rehearse restoration rather than assuming a cloud label automatically provides continuity.
