Basic concepts
A few basic concepts are necessary when you use AIDA.
- KPIs (Key Performance Indicators)
- KPIs for HCL Workload Automation
processes that are constantly monitored by AIDA. For example, the number
of completed jobs in the current plan.
For more information about HCL Workload Automation KPIs managed by AIDA, see: KPIs for HCL Workload Automation.
- Anomaly Source KPI
- The KPI whose anomalous trend has triggered an alert.
- Correlated KPI
- KPI correlated with the anomalous KPI. You can add one or more correlated KPIs to the
anomaly data analysis.
For more information about how to add correlated KPIs to the anomaly data analysis, see: Analyzing an alert instance.
- Data Point
- Each singular observation of a KPI.
- Anomaly
- Unexpected KPI data point.
AIDA detects an anomaly when a KPI falls outside the expected range of values which is statistically defined based on KPI historical data.
For more information about anomaly data analysis, see Analyzing an alert instance.
- Alert
- Sequence of anomalies of a KPI.
- Alert Instance
- A single occurrence of an alert, given its definition. As AIDA continuously monitors
KPIs, when an alert is triggered, a record is created into OpenSearch database with the alert
instance information.
For more information about alert instances, see: Overview dashboard.
- Alert Severity
-
For each detected anomaly, AIDA calculates its percent deviation from the interval estimation. When an alert is generated, given its definition, the alert severity is calculated as average of percent deviations of the anomalies that concur to the alert generation. Alert severity classification by severity is:
- High, when the average of percent deviations is > 30
- Medium, when the average of percent deviations falls in the interval 20-30
- Low, when the average of percent deviations is < 20
- Anomaly Bounds
- The upper and lower bounds of the expected range of values for a KPI.
- Alert Trigger
- Set of conditions that define an alert. For example: 10 consecutive KPI data points that
fall outside the expected range of values within 1 hour.
When triggering conditions are satisfied, a new alert instance is created inside OpenSearch database.
There are two types of alert triggers available for selection:- Continuous: Triggers when anomalous data points fall above or below the predicted range.
- Total: Triggers for anomalous data points that are either above or below the predicted range, as well as those that exceed both thresholds.
- Anomaly %
-
The percentage of observed KPI data points that fall outside the expected range of values in the reference time interval:
- < 6 : Low
- 6-10: Medium
- >10: High
- Anomaly Data Analysis
- Anomaly Data Analysis is part of AIDA User Interface. When
anomalies in a KPI trend generate an alert, you can compare the anomalous trend with trends in
one or more different time intervals. You can also add correlated KPIs to the anomaly data
analysis to find root causes faster.
For more information about anomaly data analysis, see Analyzing an alert instance.
- Alert Details
- Alert Details provides detailed information about an alert, it's status, the current opened
instances and its history.
For more information, see: Alert details.
- Alert History
- Calendar graph showing previous alert instances and related severity.
For more information, see: Alert details.
- Timerange
- How often a KPI is checked to detect anomalies (for example: every day, or every 10 minutes). It is set through the PROPHET_ORCHESTRATOR schedule_alert parameter of the common.env configuration file (or in the value.yaml file for Kubernetes deployments).
- Special Days
-
Special days are days on which a KPI trend is affected by seasonality factors such as national holidays, vacation, business cycles, recurring events. To avoid false positive alerts, the special days are included in AIDA prediction model with a higher tolerance level than the standard days.
For more information, see: Managing special days.