Diagram

Four states IDLE, REQUEST ARRIVES, SERVING PEAK, TRAFFIC STOPS in a loop. Animation moves a request dot along the arrows while each state highlights in order.
Automatic Scaling From Zero to Peak Demand
Automatic scaling from zero to peak demand means the platform adds and removes capacity in response to traffic without manual rules. In Cloud Run, each revision is automatically scaled to the number of container instances needed to handle all incoming requests. When request volume rises, additional instances start instantly to absorb the peak. When a revision receives no traffic, it is by default scaled to zero container instances.
This behavior is different from traditional autoscaling, which often requires fine-tuning thresholds and policies. Serverless autoscaling is instant and elastic, so it can move from idle with zero instances to serving with many instances and back to idle without operator action. If a workload cannot tolerate a cold start, the team can set minimum instances to keep one or more instances warm and ready for the next request. Choose scale-to-zero for variable or infrequent traffic, and choose minimum instances when fast response matters more than eliminating idle capacity.
Domain 3 of 3
27 practice questions
Practice this DomainLast updated