Severity can be:
critical, 
major, 
minor, 
warning, or 
indeterminate. Monitoring provides metric anomaly detection by leveraging IBM SmartCloud Analytics - Predictive Insight microservices. This provides real-time performance analysis; anomalies are detected when the value of a metric deviates from the metric's baseline. Use baselines to detect anomalies in the behavior of data.
A baseline is a model of typical behavior for a resource metric. Baselines are created using machine learning and are based on the learned behavior for a specified resource metric. Baselines are recorded against intervals of time.
When you create a baseline, you select metric resources for anomaly analysis - baselines are not created out of box. This begins the process of collecting the metric historic and future data to build the model that defines the normal range of values for any time interval for the resource. This requires queries against the entire set of raw data for the metrics that is being baselined. All of the data that is required for baselining will not be in memory.
Upper bound and lower bound values are dynamically established for a resource metric. Upper bound and lower bound values vary depending on the normal behavior at a given time of the day or week (intervals). They cannot be manually customized but as your data matures, the upper bound and lower bound values are dynamically updated.
You can configure an event to trigger when an upper bound and/or lower bound value is breached. Specify how many consecutive intervals can elapse with the breached upper bound and/or lower bound value before the event is raised. You can configure the
severity of an event.
Severity can be: critical, major, minor, warning, or indeterminate.
It can take up to 24 hours for the first baseline data points to start appearing in the Resources dashboard. 14 days of raw metric data is recommended to see accurate baselines. However, baselines are still generated with less than 14 days of historical data, but they will present a less accurate baseline model.
Raw metric data retention is controlled by the retentionResolutionRaw parameter in the custom resource YAML file for the observability service in Red Hat Advanced Cluster Management.
For more information about raw metric retention and how to change it, see the Enabling the observability service in Red Hat Advanced Cluster Management topic.
When you install IBM Cloud Pak for Multicloud Management, the baselines feature is enabled by default. To disable the baseline feature, set the monitoringDeploy.global.monitoring.analytics parameter to false. For more information, see the
advanced Monitoring parameters topic.
Note: You cannot create a baseline if more than 7 days of metric data exits This is a temporary limitation resulting from limitations in the Red Hat Advanced Cluster Management observability service.
Note: Data displayed in baselines is not up to date This is a temporary limitation resulting from limitations in the Red Hat Advanced Cluster Management observability service.