A team configured an Instana alert for CPU utilization exceeding 80% on a group of application servers. The alert fires 3-4 times daily, but when engineers investigate, observed CPU usage is around 45-55%. Reviewing Instana metric graphs reveals brief CPU spikes to 85-90% lasting only 2-3 seconds. What is the most appropriate configuration change to eliminate these false positive alerts?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Brief metric spikes that self-resolve in seconds are common in healthy production systems and rarely indicate actionable problems. Adding a sustained duration requirement means the alert only fires when the condition persists long enough to warrant human attention, filtering out transient activity while still catching genuine CPU saturation. Raising the threshold risks missing real saturation events, and maintenance windows would suppress legitimate alerts during business hours.
Full explanation below image
Full Explanation
Brief metric spikes that self-resolve in seconds are common in healthy production systems and rarely indicate actionable problems. Adding a sustained duration requirement means the alert only fires when the condition persists long enough to warrant human attention, filtering out transient activity while still catching genuine CPU saturation. Raising the threshold risks missing real saturation events, and maintenance windows would suppress legitimate alerts during business hours. The correct answer is 'Configure the alert to require the CPU threshold to be exceeded for a sustained minimum duration, such as 5 minutes, before the alert fires'. The incorrect options — 'Raise the CPU alert threshold from 80% to 99% to avoid triggering on brief spikes', 'Replace the threshold alert with an Instana built-in anomaly detection alert', 'Add a maintenance window during business hours to suppress the alerts' — are wrong because they do not align with IBM Instana's architecture or recommended practices for this scenario. Understanding this concept is essential for the Domain 6: Troubleshooting domain of the IBM Instana Observability certification.