AWS EKS Cluster (Kubernetes)

You'll find a real EKS cluster in your Tensor9 AWS account.

This EKS cluster high fidelity digital twin mirrors the operational state of the Kubernetes cluster running in the customer appliance. Here's what you can do:

• View Namespaces, Pods, and Deployments: Open the EKS console or use kubectl in your Tensor9 AWS account to view namespaces, pods, and deployments. These resources correspond to the Kubernetes components running inside the customer appliance.
  • Allow-Listed Access: Customers must explicitly allow-list which namespaces, pods, and deployments are accessible through the digital twin.
  • Audit Logging: All access and operations performed through the digital twin are appended to the customer’s audit log for full traceability.
• Observe Pod Logs: Access individual pod logs through the EKS console or via kubectl logs. These logs are synchronized in real time with the logs produced by the corresponding pods in the customer appliance.
  • Log Allow-List: The customer must specify which logs can be accessed through the digital twin.
  • Audit Logging: All viewed logs are recorded in the customer’s audit log.
• Monitor Metrics: Metrics such as CPU and memory usage for nodes and pods are synchronized and viewable through CloudWatch or other monitoring tools integrated with the EKS cluster.
• Debug Failures: If a pod crashes inside the customer appliance, the corresponding pod in the EKS cluster digital twin will also terminate. This allows you to detect and diagnose pod failures without needing direct access to the customer’s environment.
• Perform Actions: You can delete, restart, or scale deployments within the EKS cluster digital twin. However, all actions, such as scaling down pods, require explicit allow-listing by the customer and will be logged in the customer’s audit log.

The EKS cluster digital twin is designed to be lightweight:

• A pod requiring multiple vCPUs and high memory in the customer appliance may be represented by a pod requiring minimal resources (e.g., 0.25 vCPUs and 256 MB RAM) in the digital twin.

This keeps infrastructure costs low while preserving real-time observability and control.