Google Cloud has introduced Fault Injection Testing in preview, a service designed to test how cloud applications respond to disruptions.
The tool lets developers and architects create experiment templates that inject faults into selected cloud resources and observe how systems behave during failures. It is intended to help teams verify that safeguards and recovery processes work before a real outage affects customers.
Fault Injection Testing arrives as businesses run more distributed applications across cloud environments, where resilience testing can be harder to carry out than in self-hosted data centres. In those settings, organisations often have less direct access to the underlying infrastructure when testing failover scenarios.
That gap can create operational and regulatory risks, Google Cloud said, citing potential reputational damage from service failures, compliance concerns in regulated sectors such as finance, and delays to migration projects when teams cannot verify that critical applications will remain stable during a zone failure.
How it works
The service uses experiment templates to define both the type of fault and the resources affected. In the current preview, users can test two main scenarios: failover of a high-availability Cloud SQL instance from a primary zone to a standby zone, and degraded application traffic through a Layer 7 load balancer by adding latency and HTTP error codes.
Before any test runs, the system carries out a dry run. This read-only simulation checks permissions and produces a current list of affected resources, allowing teams to review the scope before starting the experiment.
Once reviewed, the experiment can be launched manually. The injected fault runs for the duration set in the template, after which the system automatically reverts the changes.
If a test produces unexpected results, users can stop it and begin reverting resources to their normal state immediately. During the preview period, Google Cloud recommends using the service in non-production environments.
Early users
Google Cloud named KeyBank and Servier as early users. The organisations are using native fault injection to simulate demanding failure scenarios such as zonal outages and test whether services remain stable under those conditions.
The preview is available through the Google Cloud console, the gcloud command-line tool, and REST APIs. Access requires enrolment in the preview, activation of the Fault Testing API, and assignment of the relevant operator role for teams configuring and running experiments.
For organisations adopting cloud services at scale, resilience testing has become a more visible part of operational planning. Financial institutions and other regulated industries in particular face pressure to demonstrate disaster recovery arrangements and show that essential systems can continue operating during infrastructure faults.
The launch also reflects a broader move by cloud providers to offer built-in testing tools rather than leaving customers to assemble bespoke methods for simulating outages. Native services can give platform users a more consistent way to test databases, network paths, and application traffic within managed cloud environments.
In practical terms, Google Cloud is starting with a narrow set of failure modes rather than a broad catalogue of infrastructure disruptions. The initial focus on Cloud SQL failover and load balancer traffic degradation suggests it is targeting common scenarios that affect application availability while using the preview period to gather feedback on how the service fits with existing testing practices.
Users are advised to begin by creating a template for either a Cloud SQL or load balancer resource and running a dry run in a non-production environment to assess the potential impact.