Overview of the 'anomaly detection' feature and its benefits

Products

• VMware Telco Cloud Service Assurance

Issue/Introduction

This article provides an overview of the 'Baseline and Anomaly Detection' feature in Telco Cloud Service Assurance

Environment

Telco Cloud Service Assurance 2.X

Resolution

The Anomaly score signifies the level of uncertainty in that metric value. If the probability of metric value is higher, then the anomaly score of the metric value is low.
If the probability of the metric value is lower, then the anomaly score of the metric value ls higher.

• UCB signifies the upper threshold of the metrics value based on the previous time interval data.
  The UCB is sum of mean of pervious time interval 's data and standard deviation of previous time interval's data scaled by confidence interval (User configurable | Default = 2)
• LCB signifies the lower threshold of the metrics value based on the previous time interval data.
  The LCB is sum of mean of pervious time interval's data and standard deviation of previous time interval's data scaled by negative confidence interval (User configurable | Default = 2).

Anomaly score calculation takes place, if the metric value is higher than UCB or lower than LCB.
In other words, detection of anomaly is not based on score, rather based on the UCB or the LCB.

Hence, if the metric value in question is found to be anomalous, only then score is calculated.

Additional Information

More details about Configuring Anomalies is available at: Configuring Anomalies