To avoid alert fatigue from brief latency spikes, which configuration change in a Cloud Monitoring alerting policy prevents notifications for transient blips?

The 'for' field in a Cloud Monitoring alerting policy defines the duration a metric must continuously violate the threshold before an alert is triggered. By setting this duration, you ensure that only sustained issues generate notifications, effectively filtering out transient spikes and reducing alert fatigue. Email-only notification channels do not inherently batch short alerts; they simply change the delivery method. Lowering the alert threshold would increase, not decrease, the number of alerts, requiring more manual dismissal. Increasing metric collection frequency would provide more granular data, potentially making transient spikes more visible and thus more likely to trigger alerts if not combined with a duration-based condition.