The OpenSearch Anomaly Detection plugin enables identification of anomalies in time-series data. Anomalies are considered outliers or non-standard data points with regards to the trends in your time-series data. Using the Anomaly Detection plugin, indices you specify are analysed by the Random Cut Forest (RCF) unsupervised machine learning algorithm to build a model of your time-series data to detect anomalous data points on either historical data or new data ingested into the target indices.
Enabling the Plugin
When choosing the eligible OpenSearch version (OpenSearch 1.3.7 and 2.4.0 onwards respectively), the options will be available on the Console and the API to provision clusters with the Anomaly Detection plugin. Existing clusters satisfying these version requirements can also have the Anomaly Detection plugin enabled through making a request to support.
Via the Instaclustr Console, you can enable the Anomaly Detection plugin within the OpenSearch Setup step when creating a new cluster.
API and Terraform
Support for API and Terraform V1 provisioning are also available, in order to utilise these provisioning routes, see the Terraform Provider repository and API Documentation respectively.
The Anomaly Detection plugin can be used via either the REST API or through Dashboards through the left hand sidebar. For further information on the steps to create and use anomaly detectors, more information can be found on the official OpenSearch documentation