Summary
- A power event affected Google's europe-west2-b availability zone on 20 October, with errors also reported across parts of the London region.
- Google's public incident record covers seven hours and 52 minutes, from 07:08 to 15:00 BST, although individual customer recovery times varied.
- Financial services, delivery businesses and courts experienced disruption; the precise electrical failure mechanism remains undisclosed.
Google Cloud has closed an incident affecting its London cloud region after a power event disrupted services in the europe-west2-b availability zone on 20 October, with related errors reported across parts of the wider region.
Google’s service health record gives an incident duration of seven hours and 52 minutes. The disruption began at 06:08 UTC, or 07:08 BST, and the recorded period ended at 14:00 UTC, or 15:00 BST. Recovery was progressive rather than an instantaneous return of every customer workload.
Google attributed the initial outage to a power event and reported that the electrical problem had been mitigated while engineers continued restoring services. The provider has not published a completed root-cause analysis identifying the failed equipment or explaining every dependency involved in the regional disruption.
Customers encountered unavailable zonal services, elevated errors and intermittent network problems. Financial institutions, delivery businesses and public services reported difficulties during the morning, although the precise impact on each organisation depended on its own systems and dependencies.
A zonal failure spread into regional services
Cloud regions contain separate availability zones intended to provide locations from which customers can design services that tolerate infrastructure failures. Such separation does not guarantee that applications operate continuously when one location becomes unavailable.
Google described services in europe-west2-b as unavailable during the incident, with some services elsewhere in europe-west2 experiencing elevated error rates and network issues. The distinction matters because the source identified a particular affected zone while acknowledging wider service consequences.
An application may depend on computing instances, storage, databases, identity services and network components with different geographic configurations. Replicating one element without accounting for the others can leave a service unable to function when a dependency fails.
Customers can design workloads across several zones or regions, but doing so involves extra resources and arrangements for data consistency, routing and recovery. The presence of multiple availability zones in a provider’s region is not evidence that an individual customer has implemented them.
During recovery, Google suggested that customers could attempt to migrate affected workloads to another available location where feasible. It also reported difficulty creating new workloads while engineers prioritised existing ones, limiting the usefulness of migration for some affected users.
Moving a workload requires capacity at the destination and suitable data and connectivity arrangements. Stateful applications or those tied to particular zonal resources may require additional recovery work, even after the underlying electrical event has been contained.
Banking and public services reported disruption
Revolut reported problems affecting services including card payments and transfers while assuring customers that funds and data remained safe. Lloyds Banking Group brands, HSBC and other financial services also experienced difficulties as the disruption became visible to users.
Problems were reported at organisations beyond banking, including delivery and property services. Britain’s courts and tribunals experienced interruptions to remote hearing infrastructure, demonstrating the operational consequences when business-critical applications depend on external cloud capacity.
Not every incident reported during the same hours can automatically be traced to Google. The cloud provider’s dashboard establishes the regional infrastructure event; individual service providers are responsible for confirming their own dependencies and assessing how the event affected their applications.
The Financial Conduct Authority said it was examining the consequences with affected financial firms. Service outages at important infrastructure providers create questions about recovery arrangements, contractual responsibilities and how institutions validate their continuity planning.
Google did not publish a list of all end customers affected or a uniform downtime figure for their services. The dashboard’s seven-hour-and-52-minute duration describes the provider’s incident record, not necessarily the interruption experienced by each bank, court or business.
Electrical restoration is only part of recovery
Data centre electrical systems typically include distribution equipment, uninterruptible power supplies and alternative supply arrangements designed to protect services against defined faults and interruptions. Their effectiveness depends on the actual configuration and the equipment affected by an incident.
Google has not said whether the London event originated in incoming utility supply, switchgear, a transformer, an uninterruptible power system or another component. It would be unsupported to conclude that a particular backup system failed or that the event involved sabotage.
Even when electricity is restored, computing services may need additional checks and recovery steps. Equipment, storage and distributed applications can have different restart dependencies, while operators must verify that workloads can operate reliably before declaring the incident resolved.
Google’s updates recorded mitigation of the power event before the restoration of all affected workloads, consistent with the separation between an electrical fault and its downstream service consequences. They do not provide the detailed technical sequence needed to assign a definitive root cause.
The London incident is now closed in the public service health record. A fuller investigation will be required to determine the electrical failure mechanism, the path by which disruptions affected other regional services and any resulting changes to resilience arrangements.
