An incident is an episode characterized by abnormal performance, availability, or web traffic volume as compared to a baseline. Examples of incidents include:
The system prioritizes incidents by type of abnormality, duration, and the number of users affected.
To detect incidents, set up your system to monitor the appropriate parameters and take the appropriate measurements, as described in the following topics:
The system can monitor Watchpoints, SLTs, and error conditions for incidents:
The system does not support monitoring of informational conditions for incidents.
Abnormality describes how exceptional or atypical an incident is.
The system measures abnormality based on a logarithmic scale. It establishes a baseline for normal trends in your web traffic, and then calculates the number of standard deviations from the baseline that your web traffic exhibits during a 30-minute period.
The number of standard deviations away from the baseline is the incident's abnormality. Each type of incident has its own type of abnormality, as described in the following table:
Type of abnormality
The system detected an abnormally high percentage of requests with errors for a particular Watchpoint.
The system detected an abnormally high percentage of pages that violated their SLTs for a particular Watchpoint.
The system detected an abnormally high or abnormally low amount of traffic on a particular Watchpoint.
By default, the system represents abnormality as an index value from 0 to 4.
Abnormality rating based on multiples of the standard deviation
Number of standard deviations from baseline
Less than or equal to 1
1 – 2
2 – 4
4 – 16
Greater than or equal to 16
For each type of incident, the system calculates the number of user sessions affected (the number of distinct sessions detected for the Watchpoint during the specified period when the incident occurred).
To monitor incidents that occur in your web traffic, you can:
Watch the Incidents page of a Real User Analyzer component, which provides the list of incidents occurred during the specified period of time
Monitor the incident-related dashlets on the Real User Analyzer dashboard