How to Meter Execution DurationMeasuring CPU Hours of containers, pods, and serverless in seconds

Cloud and DevOps companies often charge customers based on execution duration, such as CPU hours. However, metering workloads can be complex due to challenges in detecting start and stop events, handling lost events, long-running workloads, and shifting billing periods. An alternative approach is heartbeat style metering, which involves periodically checking if a workload is running and incrementing a counter associated with the workload. This method reduces the possibility of under or overbilling due to missing lifecycle events and offers more accurate usage tracking. However, it generates a high volume of usage events that may require streaming aggregation for efficient processing.