Alerts are managed on the Alerts > Alert configuration page. When you open it and click on a server, instance or database node, you should see something like this:
Here, the selected category is CPU, so all the CPU-related performance counters for the selected node are listed. For each counter, there are two slider controls, five checkboxes and a numeric field.
The slider controls are used to set the alert and warning thresholds. Here, the alert threshold for CPU USAGE (%) is set to 80, which means that any data point that exceeds it will raise an alert.
By checking one of the email checkboxes, you can receive emails about warnings or alerts. For more information on configuring emails, see the article How to receive emails about problem situations.
Checking the checkbox with the exclamation point will mark the counter as critical. Critical alerts are always prioritized in the alerts section of the Dashboard.
The last three checkboxes are used to enable linear and anomaly detection alerts. To learn more about predictive alerts, see the article What are predictive alerts?.
The numeric field is used to set the limit for the counter. The limit determines how many consecutive data points over a threshold are needed before an actual alert is raised. A limit of 1 means that every time the counter is above the threshold, an alert is raised. For more information, see What does the "limit" setting do?
You can also let SQL Governor auto-adjust your alerts and warnings. To learn how to do that, see the article How to auto-adjust alerts and warnings.
If you click on a database node and then right-click over it, you can set the database to collect query anomaly predictions. Only one database can be active at a given time.
Query anomaly predictions are shown on the dashboard.