> [!WARNING]
>
> **Preview**
>
>
> This product is
>
> subject to the "Pre-GA Offerings Terms" in the General Service Terms section of the
> [Service Specific
> Terms](https://docs.cloud.google.com/terms/service-terms#1).
>
> Pre-GA products are available "as is" and might have limited support.
>
> For more information, see the
> [launch stage descriptions](https://cloud.google.com/products/#product-launch-stages).

This page links to instructions for creating an experiment for specific faults.
An experiment is an instance of a fault injection experiment against your target
resources, which should fail because of the fault.

To set up a new experiment, you'll create an experiment template, which defines
the fault to be injected and the target resources. Choose from the following
pages based on the fault to inject.

- [Fail Compute](https://docs.cloud.google.com/fault-injection-testing/experiment-setup/fail-compute.md.txt) - simulate the unavailability of GKE or Compute Engine [compute](https://docs.cloud.google.com/compute/docs/overview.md.txt) resources.
- [Degrade Application Traffic](https://docs.cloud.google.com/fault-injection-testing/experiment-setup/application-traffic.md.txt) - simulate traffic degradation through a layer 7 [load balancer](https://docs.cloud.google.com/load-balancing/docs/load-balancing-overview.md.txt).
- [Failover Cloud SQL](https://docs.cloud.google.com/fault-injection-testing/experiment-setup/cloud-sql.md.txt) - fail an HA-enabled [Cloud SQL](https://docs.cloud.google.com/sql/docs/mysql/high-availability.md.txt) database from primary to standby.

These faults are designed to restrict the scope of the experiment to a single
project's resources in a single region or narrower. The experiment won't affect
a wider scope than what is intended. However, these faults cause real failures
to the targeted resources in its environment.

By targeting infrastructure components such as a Cloud SQL database or an
Application Load Balancer, as well as other components that will be supported in
the future, you can reasonably approximate a zone or region failure in the
context of your own application.