Google adds native scale-to-zero support to GKE

Google has added native scale-to-zero support to Google Kubernetes Engine 1.37, allowing idle workloads to stop all their running containers and restart when demand returns. The feature uses external demand signals to tell the autoscaler when to stop or wake a workload, removing the need for an optional scaling component. Google says shared capacity buffers can provide resources for a restarting workload without waiting for a new node. Those buffers retain warm computing capacity, so the faster restart option still requires resources to be kept available.