Infrastructure Resilience Briefing — April 26, 2023 Google Cloud Paris Cooling Incident
Cooling system and water leaks in Google Cloud’s europe-west9 (Paris) region forced compute and storage shutdowns; Google’s status update outlines gradual recovery and credits through the general GCP SLA for affected services.
Executive briefing: A cooling subsystem failure and water intrusion in Google Cloud’s europe-west9 region on triggered shutdowns of multiple data halls, leaving Compute Engine and Persistent Disk volumes unavailable for several hours while engineers restored capacity.1 Google advised customers to fail over to alternate regions and noted that standard SLA credits would apply for workloads that breached availability targets.2
Regional impact
- Compute disruption: Virtual machines in the impacted data halls were preemptively shut down to protect hardware, producing VM unavailability and delayed restarts.1
- Storage risk: Some Persistent Disk replicas became temporarily unavailable, prompting Google to recommend restoring from backups or failing over to unaffected zones.
- Recovery staging: Google restored capacity in phases over several days while monitoring humidity and electrical safety, creating prolonged failover windows for customers.
SLA and credit posture
- Map downtime windows against the availability targets in the GCP Service Specific Terms and submit claims within the reporting deadline.2
- Document when workloads were shifted to alternate regions, capturing evidence for resilience KPIs and future credit claims.
- Prioritize multi-region backups for Persistent Disk and Cloud SQL instances that remain zonally deployed in europe-west9.