Anomaly_Detection.md

Anomaly Detection

Source: https://docs.rubrik.com/en-us/saas/saas/anomaly_detection.html

---

Anomaly Detection

Anomaly Detection examines a range of file system statistics to check for anomalies in file system usage patterns.

Anomaly Detection provides details about potentially anomalous incidents, including their location, Rubrik cluster, file details, and snapshot time. When you enable Orchestrated Recovery, objects that are part of a Recovery Plan are grouped together.

When you enable Anomaly Detection for the data, Large Language Model (LLM) analysis for Anomaly Detection is enabled by default. When LLM analysis for Anomaly Detection is enabled, RSC uses a pre-trained LLM in Microsoft Azure OpenAI to detect suspicious activity. The model differentiates suspicious activity from legitimate changes caused by routine operations, such as upgrading an application or operating system.

Identifying and differentiating malicious behavior from legitimate activities helps reduce the frequency of false alerts and enhances the experience of using Anomaly Detection to protect data. When you do not want to use LLM analysis on Anomaly Detection, you can disable it and use Anomaly Detection without it.

When you enable Data Security Posture for an object that has an anomalous snapshot and a user account that has Data Security Posture privileges, the Sensitive files and Sensitive hits columns display values based on the base snapshot of the anomalous snapshot. The Sensitive files column shows the number of files that contain sensitive data, and the Sensitive hits column shows the number of sensitive hits for that object.

Note:

Anomaly Detection supports protected objects running in single-stack IPv6 environments.

On the Anomaly Detection page, you can:

• View the list of new and resolved investigations.
• Use the filters on the page to sort anomalies by time range, severity, encryption, object type, Rubrik cluster, and SLA Domain.
• Sort anomalous data by sensitive files or sensitive hits when you enable Data Security Posture.
• Filter anomalies by the selected policies when you enable Data Security Posture.
• Investigate and recover anomalous files in protected workloads.
• Recover a full snapshot for a virtual machine.
• Use the Search filter to search for protected workloads or Recovery Plans that have an identified anomaly.
• Initiate bulk recovery for vSphere-protected workloads or for Recovery Plans that have an identified anomaly.
• Perform cyber recovery on virtual machines based on threat hunting results to a local, isolated recovery environment.
• Download a CSV file of the changes identified for each protected workload. When you click Download CSV, RSC prepares a secure link for the file and displays a message about the download. You can also visit the File Preparation Center to access the downloaded file.

RSC supports Anomaly Detection for:

• VMware, AHV, and Hyper-V virtual machines
• NAS filesets
• Linux and Windows servers
• Windows Volume groups
• NAS CD data sets
• Azure virtual machines
• Azure Blob storage
• Microsoft OneDrive and SharePoint
• Amazon S3 buckets
• Amazon EC2 instances and EBS volumes
• Kubernetes virtual machines
• Kubernetes protection sets
• GCP GCE Instance
• GCP Persistent Disk