Techniques described herein automatically check for persistently inactive instances, based on defined metrics, and auto-archive such instances to lower-cost cloud resources. An inactivity time threshold is dynamically adjustable to a longer or shorter time period based on current load running on limited/more expensive resources to more aggressively or less aggressively archive the inactive instances, thus enabling additional active instances to run on the limited/more expensive resources and supporting more total users.
展开▼