Google Cloud's London power fix leaves service recovery as the next test
Google reports that the power event behind a London cloud outage is mitigated. Its service notice shows why infrastructure repair and customer recovery need separate checks.
When a cloud provider fixes a power problem, can a customer get back to work? Google Cloud's London incident offers a practical reason to check both milestones before calling an outage over.
In its official incident notice, Google's October 20 update at 7:12 a.m. Eastern says the power event affecting the europe-west2-b zone has been mitigated. The same update still describes Google Cloud services as unavailable in that zone, with some services across the wider europe-west2 region potentially experiencing elevated errors and intermittent network problems.
The dashboard identifies the affected location as London. It dates the incident's start to October 19 at 11:08 p.m. Pacific, which is October 20 at 2:08 a.m. Eastern. Those are two time zones describing the same start, not separate incidents.
The recovery checkpoint is the usable workload
Google's 7:12 a.m. update anticipates partial infrastructure mitigation within 30 minutes, followed by the remaining services. That is the provider's recovery expectation at that update, not confirmation that those steps have happened. The inspected notice does not give a completed service-restoration time.
Our reading is straightforward: fixing the initiating power event and restoring a customer's application answer different operational questions. A customer needs the services its workload depends on to function together. An infrastructure milestone alone does not demonstrate that a transaction can finish or an application can serve users again.
The distinction also matters when assessing cloud businesses. Reliability can involve both the provider's recovery work and a customer's spending on resilience. The useful business question is what it costs to maintain an acceptable service through a disruption, and how much recovery work the customer must perform. This incident notice does not quantify either cost.
Failover advice comes with a condition
Google suggests that affected customers consider moving impacted workloads to another available zone or region if feasible. That qualification matters. An available destination does not, by itself, establish that a particular application can move there successfully.
For a customer evaluating its own recovery plan, useful checks include whether required data and dependencies are available at the destination and whether the recovered application actually works. These are analytical questions raised by the provider's advice, not findings about a named customer's systems.
The next evidence to watch is confirmed restoration of the affected services and a fuller account of the incident. This notice establishes Google's reported operational impact; it does not establish a global outage, identify affected customers or measure lost revenue, compensation or financial damage. Those conclusions require additional evidence.
Source cutoff: Google's October 20, 2026 update at 4:12 a.m. Pacific / 7:12 a.m. Eastern. Incident status may change after publication.