Introducing Google Cloud Fault Injection Testing (FIT) in preview

Google Cloud Fault Injection Testing (FIT) is now available in preview to address challenges in maintaining high availability for mission-critical cloud services. The tool allows teams to deliberately introduce faults—such as zonal outages or latency spikes—into their environments to verify failover mechanisms before real disruptions occur. This approach helps identify vulnerabilities in safety protocols that could otherwise lead to prolonged outages or performance degradation. By automating these tests, FIT aims to reduce the manual effort required to validate system resilience in complex distributed architectures.

FIT operates through experiment templates that define specific faults and target resources, such as triggering a failover for a high-availability Cloud SQL instance or degrading application traffic via a Layer 7 load balancer. Before executing any fault injection, the service performs an automated dry run to check permissions and list affected resources, ensuring transparency. Users can manually start the experiment, monitor behavior, and revert changes if issues arise, with faults automatically reverted after a predefined duration. This controlled approach enables teams to validate application stability without risking customer-facing disruptions.

During the preview phase, Google recommends using FIT in non-production environments to assess its integration with existing testing workflows and provide feedback for product improvements. Early adopters like KeyBank and Servier are leveraging the tool to simulate demanding failure scenarios, such as zonal outages, to strengthen their service stability. The preview period serves as an opportunity for organizations to refine their reliability strategies before broader deployment.

Access to Cloud FIT requires enabling the Fault Testing API in the Google Cloud console and assigning the roles/faulttesting.operator role to team members. Users can then create templates for resources like Cloud SQL or load balancers, execute dry runs, and run experiments to evaluate system behavior under simulated faults. For implementation guidance, organizations are advised to consult their Google Cloud Account Team or refer to the official User Guide for FIT.