Google to Engineers: “It’s Time to Make Feature Flags Mandatory”

A global outage of Google Cloud on June 12, 2025, caused by a null pointer exception in a core backend service, highlighted the critical importance of feature flags for ensuring system reliability and minimizing downtime. The incident resulted in widespread disruptions, affecting services such as BigQuery and Cloud Run, and underscored the limitations of relying solely on DevOps practices, which successfully deployed the code but lacked runtime control to mitigate the issue quickly. Google’s postmortem emphasized that the absence of a feature flag was a pivotal factor, preventing the issue from being caught in staging. This has led to a push for adopting "FeatureOps" as a necessary extension of DevOps, advocating for mandatory feature flag coverage on all critical binaries with a default-off posture to enhance runtime safety and prevent similar catastrophic failures. The incident serves as a call to action for organizations to enforce feature flags across both backend and frontend changes, treating them as essential infrastructure for maintaining stability and protecting against costly outages.