Anomaly-Based Alertingrepresents a shift from static thresholds to behavioral analysis in the HCO platform.
According to theSolarWinds HCO Alerting Enginedocumentation, this feature uses machine learning to establish a "baseline" for specific metrics like CPU load or memory usage over a period of 7 to 30 days.
The primary benefit is that itanalyzes entity behaviorand triggers an alert only when a metric deviates significantly from its historical "normal" for that specific day and time. For example, if a server traditionally runs at 90% CPU during a Sunday night backup, a static 80% threshold alert would trigger a "false positive" every week. Anomaly-based alerting learns this behavior and will only fire an alert if the CPU hits 90% on a Tuesday morning when the normal load is only 20%.
This reduces alert noise by focusing ontrue anomaliesrather than simple threshold violations. It does not
"remove the requirement for trigger conditions" (Options B and C); instead, it replaces a static numerical threshold with a dynamic, machine-learned threshold. The administrator still defineswhichentities to monitor andhowsensitive the anomaly detection should be.