You can specify the following parameters for Application Check metrics:
Collection Frequency
For Application Check metrics of type HTTP Health API and Prometheus, the collection frequency is the time interval (in minutes, hours, or days) at which a Simple Diagnostics Agent measures the metric. For example, if you set a collection frequency of one minute, the Simple Diagnostics Agent measures the metric once each minute.
Thresholds
A threshold defines when a metric will receive a certain rating, such as Info (blue), Ok (green), Warning (yellow), Critical (red), or Fatal (red). If no threshold is defined for a metric, Health Monitoring uses the rating for the metric reported by the cloud service.
The types of threshold available depend on the type of metric:
The following threshold types are available:
Example of a Performance metric
The rating of the metric cpu.usage is set to Warning when the value of the metric is above 50%. The rating of the metric cpu.usage is set to Critical when the value of the metric is above 80%.
Example of a metric of the type Quota
The rating of the metric disk.usage is set to Warning when the usage is above 50%. The rating of the metric disk.usage is set to Critical when the usage is above 90%.
Alerts (optional)
When you don't maintain a threshold, an alert is generated when the cloud service reports a metric status red (critical or fatal). When you maintain a threshold for a metric, an alert is generated when the defined threshold is reached. Alerts are displayed in the Health Monitoring and Alert Management applications.
The following Alert settings are available:
Alert Rules
An alert rule selects metrics of the destination. Select metrics by either their name or their label.
Use the Include table to select metrics for which alerts are created.
Use the Exclude table to select metrics for which no alerts are created.
Availability Alerts Generated from Application Check Metrics
Availability alerts can be generated from Application Check metrics (HTTP Health API and Prometheus). A metric is reported to Service Availability Management as an unplanned outage if the following prerequisites are met:
An active service definition has been created for the technical system in Service Availability Management as explained here.