Google Cloud’s partial outage in us-central1-b this week had an unusually physical root cause: network fiber cables were unplugged during maintenance, according to The Register, citing Google’s service health report. The incident ran from 07:41 to 11:52 PT on September 1. Google described the impact as “severe network degradation and resource isolation” for a portion of the affected area, according to The Register’s account of the report. At the peak, traffic flow drop rates for hosted resources reached 100%, leaving virtual machines unreachable and producing elevated packet loss. The important detail is not simply that a cable was removed. Google’s preliminary root cause, as reported by The Register, was that a routine hardware maintenance procedure inadvertently disconnected network fiber-optic cables. A procedural error then caused the maintenance action to sequentially unplug 100% of fiber paths across all devices within 13 minutes. That sequence mattered because Google’s data-center networking is designed around redundancy. The service health explanation, as quoted by The Register, says the environment uses multiple routing devices, physically separated fiber paths, and diverse power sources. It is intended to tolerate all single-device or single-fiber-path failures, and most double- or triple-failure cases, without affecting customer traffic. This incident defeated that design by turning a redundant system into a correlated failure. Google said the speed and nature of the error meant warnings about the incorrect action did not reach the engineer before the disconnection was complete, according to The Register. Once the relevant routers were disconnected, virtual machines in the affected zone could still communicate locally, but users could not reach them from outside. Google’s recovery path had two parts. The company moved traffic away from the disconnected routers, allowing resilience mechanisms to shift traffic to healthy capacity elsewhere in the region. Technicians then identified the disconnected optical links and reseated the fibers; after the physical links were restored, traffic flow rates normalized and capacity was returned to service. The episode is a reminder that hyperscale reliability is not only a function of redundant hardware. It also depends on maintenance procedures, change sequencing, and safeguards that can stop a human action before it converts independent backup paths into a single shared failure. Who benefits: Infrastructure teams get a concrete failure mode to review against their own maintenance runbooks and warning systems. Customers with architectures that avoid reliance on a single affected zone are better positioned when this kind of localized isolation occurs. Who's exposed: Resources hosted in the affected area were exposed to full traffic drop rates at the incident’s peak, according to Google’s report as cited by The Register. The provided reporting does not quantify customer count, revenue impact, or affected services beyond unreachable virtual machines and packet loss.