Systems and methods that can vary a data collection rate via a rate adjustment component, to collect data with different level of granularity. The rate adjustment component can further include an estimation component that can automatically predict a required sampling rate for a stage of an operation, based on statistical models and data collected for similar operations and/or history data. Such difference in the granularity level can initiate in part in response to fault detection, alert triggering, and the like. Accordingly, future trouble shooting efforts can be performed with respect to data that is typically collected at an adjustable rate.