AIOps 2.6.1 introduces a new alert surge detection option based on surge detection rules. ITOps admins with expert knowledge of their systems can set up multiple surge rules, to ensure a sudden spike in alerts is handled and alert noise reduced.
This method is based on count of alerts that satisfy ‘surge detection rules’ in each batch of alerts received in AIOps.
How to set up surge detection rule?
In Configurations -> Settings -> Surge Handling, opt for ‘Rule based surge handling’ and set up the surge rules.
For each surge detection rule, specify the following parameters
- Surge rule – using alert attributes and operators
- Surge start threshold – If no. of alerts in single batch which satisfy a rule is >= surge start, then surge begins
- Surge end threshold – If the no. of alerts in single batch which satisfy a rule is less than surge end and greater than 0, then surge is considered as ended.
- Surge active time – If there aren’t any alerts matching surge rule for more than ‘x’ minutes, then surge is considered as ended.
What happens when surge is detected?
When surge is detected, all alerts satisfying a specific surge detection rule go to same cluster and is ticketed. There can be multiple surge clusters. Every new surge detected will create a new cluster and therefore a new ticket. Regular alerts will not be correlated into surge clusters.
|