Identify whether incidents are common or Outlier to expedite incident handling.
A Outlier incident is one that has not occurred within a specific past time period. When an incident is identified as Outlier, incident responders should be particularly vigilant. This is crucial for OnCall engineers, as the impact of Outlier incidents may be unknown, and response procedures might require ad-hoc decisions. Some engineers may specifically focus on these Outlier incidents to establish standardized procedures or SOPs for handling them.This feature is currently in beta and is only available in Professional and higher subscription plans. Please contact us if you need to activate this feature.
Viewing Outlier Incidents#
Console#
1.
In the incident list page, Outlier incidents will be clearly marked with a Outlier indicator;
2.
In the incident details page, Outlier incidents will display a prominent Outlier indicator and explanation at the top.
Taking Feishu/Lark as an example, when an incident is identified as Outlier, the system will enhance the notification:
How It Works#
The system uses machine learning models to determine the similarity between incidents. When the similarity is greater than 90%, we consider two incidents to be similar.When determining similarity, we primarily consider the following factors:2.
Incident detailed description
3.
Affected services (typically extracted from service labels)
4.
Alert objects within the incident (typically extracted from resource labels)
FAQ#
Does the Outlier incident feature need to be enabled separately?
No setup or activation is required. Outlier incident identification is automatically enabled for all Professional and higher subscription plans. 