Google Cloud Blog
Follow
Simplify your resilience testing strategy with Google Cloud Fault Injection Testing
Google Cloud has introduced Fault Injection Testing (FIT) in preview to address the growing complexity of ensuring high availability in distributed systems. FIT allows developers and architects to automate failure testing by deliberately introducing faults into their cloud environment. This proactive approach helps verify safety mechanisms before actual disruptions impact customers. Native resilience testing is crucial in cloud environments due to limited direct access to underlying infrastructure compared to self-hosted data centers. Failing to prove an application's ability to survive failures risks damaged trust, compliance issues, and migration delays. FIT operates by creating experiment templates that define faults and targeted resources. Currently, it supports failover testing for Cloud SQL instances and degrading application traffic by introducing latency and HTTP errors via a Layer 7 load balancer. Before injecting faults, FIT performs an automated dry run to check permissions and identify affected resources. Users can then manually initiate the fault injection, which automatically reverts after a defined duration. During an experiment, users can verify application behavior and have the option to stop and revert faults immediately if issues arise. Google recommends using FIT in non-production environments during the preview phase for early access and feedback. Enterprise partners like KeyBank and Servier are already utilizing FIT to validate their deployments by simulating failure scenarios such as zonal outages. Getting started involves requesting preview access, enabling the Fault Testing API, assigning appropriate roles, and running a dry run in a non-production environment.