What CPU usage percentage over 5 minutes should trigger an incident?

In the context of monitoring system performance and ensuring optimal operation of Tableau Server, setting a threshold for CPU usage is crucial. A CPU usage percentage of 80% over a 5-minute period is significant enough to warrant concern and trigger an incident. When CPU usage is consistently at this level, it can lead to performance degradation, impact response times, and potentially cause failures in processing requests or running Tableau workbooks effectively.

Using 80% as a threshold strikes a balance between normal operational usage and the point where additional strain on the system could be detrimental. This level indicates that resources are being heavily utilized and may soon approach maximum capacity, thus requiring immediate investigation and action to prevent outages or slowdowns.

In contrast, lower thresholds, such as 70% or 75%, may not accurately reflect an impending crisis, as many systems can operate sustainably at these levels without struggling. Meanwhile, setting the threshold at 90% could be too reactive, allowing for too much leeway and possibly delaying necessary intervention until a critical state has already been reached. Therefore, an incident should be triggered when CPU usage reaches 80% over a sustained period, making it the most appropriate choice for monitoring system health in this scenario.