GCP Professional Data Engineer Practice Question

The metric dataflow.googleapis.com/job/system_lag measures, in seconds, the difference between the current event-time watermark and the present clock time for each running Dataflow job. Alerting on the MAX of this metric across all workers ensures the policy triggers whenever any part of the pipeline lags beyond the 5-minute threshold. Using element counts or processed-bytes metrics does not directly indicate freshness, and Pub/Sub backlog reflects only input subscription state, not end-to-end pipeline latency. Averaging system lag can mask localized hotspots because a few timely workers may dilute a rising lag on others.