Abstract: Methods and systems for deriving metrics for service and/or computer (e.g., server) utilization and stability from instrumentation of applications showing Poisson characteristics are disclosed. Such methods and systems may be further applied to automatically act upon the conditions indicated by the derived metrics, for example, to provision additional virtual server resource(s) in a cloud-based system.