Google Cloud outage exposes datacentre resilience concerns

Disruption in Netherlands renews questions about the physical resilience of cloud infrastructure

A power and cooling failure at one of Google's datacentres in the Netherlands left several Google Cloud services unavailable for almost 15 hours last week.

According to Google's incident report, the disruption affected its Europe-West4 availability zone for 14 hours and 55 minutes, beginning at 16:39 PST on 15th July and ending at 07:34 PST on 16th July.

Services impacted included Google Cloud VMware Engine (GCVE), Bare Metal Solution (BMS) and Google Cloud NetApp Volumes.

Google said the outage began after an electrical fault on the utility grid supplying the facility interrupted power distribution systems and disabled cooling infrastructure.

The company said the loss of cooling caused temperatures inside affected data halls to rise rapidly, forcing it to shut down host servers, storage clusters and network switches to prevent hardware damage.

"The loss of cooling infrastructure resulted in ambient temperatures rising rapidly within the affected data halls," the company said in its incident report, adding that systems were powered down as a protective measure against extreme heat.

Google operates several datacentres across the Netherlands, including facilities in Eemshaven and Middenmeer, as well as its newest site in Winschoten, which opened in late 2025.

The latest disruption follows a separate incident in June, when a fire at a third-party datacentre affected Google's cloud services in India.

Cloud transparency

Analysts said the disruption has renewed questions about the physical resilience of cloud infrastructure and the assumptions customers make about built-in redundancy.

Biswajeet Mahapatra, principal analyst at Forrester, told The Register customers are often encouraged to spread workloads across multiple availability zones but have limited visibility into whether some managed services still depend on a single physical facility.

He said many organisations assume cloud providers deliver greater facility-level redundancy than may actually exist for specialised services.

This architecture is not unique to Google, according to Mahapatra.

Amazon Web Services (AWS), Microsoft Azure and Google Cloud all operate some services that rely on dedicated hardware or tightly integrated infrastructure that cannot always be distributed across multiple facilities.

Adrian Wong, director analyst at Gartner, said it remains difficult for customers to understand the underlying design of individual cloud regions.

He pointed to Google's 2023 outage in its europe-west9-a region, where a water leak in part of a third-party facility caused widespread disruption.

Although Google replicates data using its Spanner database technology, one building becoming unavailable exposed limitations in that deployment.

"Our customers are often surprised by that," Wong said, referring to the lack of visibility into regional infrastructure design.

Series of infrastructure incidents

The latest outage adds to a growing list of cloud service disruptions linked to failures in physical infrastructure.

Earlier this year, Amazon reported outages affecting two of its three AWS Middle East availability zones after a fire at a datacentre in the United Arab Emirates. Its facility in Bahrain was reportedly struck again this week.

Google has also previously experienced cooling-related failures.

During the record UK heatwave in 2022, the company temporarily shut down its London datacentre after cooling systems malfunctioned under extreme temperatures.