Ataccama ONE Agentic ONE Desktop ONE DQ&C ONE MDM ONE RDM ONE Runtime Server
Downloads

Kubernetes Edge

The Kubernetes edge runs the edge compute on an Azure Kubernetes Service (AKS) cluster you operate, installed with a Helm chart. It’s the Azure counterpart to the AWS deployment options: Ataccama provides the chart, the container images, and the upgrade releases, and you run the deployment and lifecycle. For the full split of responsibilities, see Additional responsibilities for Kubernetes edge deployments.

Choose the Kubernetes edge when your sensitive data lives in Azure and you want to operate the edge on your own Kubernetes cluster. To make sure it fits your use case, review Kubernetes Edge Limitations before you decide.

How the Kubernetes edge works

The Kubernetes edge runs four workloads in your cluster:

  • edge-plane-manager: Pulls tasks from Ataccama ONE and starts one Kubernetes Job per task.

  • Job pods: Run the processing tasks and connect to your data sources.

  • edge-interactive-jobs: Handles connection tests, data browsing, and previews.

  • dq-results-reader: Returns stored results to ONE.

All traffic from the edge is outbound. Connections to your data sources go directly from the pods through your own network.

Traffic to the platform goes through the Ataccama Edge gateway, which you reach over Azure Private Link. Sign-in to the Ataccama identity provider (IDP) uses the gateway only if you enable it.

For the full list of connections, see Kubernetes Edge Network Access Requirements.

Where your data lives

  • In your Azure Blob Storage container: Profiling and data quality results (except aggregates), invalid samples, and job requests. Job requests contain connection details and queries.

  • In your Azure Key Vault, if you keep credentials there: Data source credentials.

  • In Ataccama Cloud, encrypted with a customer-specific key: Catalog metadata, data quality metrics, and profiling aggregates.

  • In Ataccama Cloud: Reference data and the results of interactive operations, such as browsing and on-demand export of values.

When you view stored results in ONE, they’re read from your container, decrypted if you enabled result encryption, and sent to the platform over TLS.

For how the Kubernetes edge fits into the overall edge architecture, see Edge Architecture.

Deploy the Kubernetes edge

Deploying the Kubernetes edge takes these steps:

  1. Check that your cluster meets the requirements in Kubernetes Edge Supported Platforms.

  2. Create the Azure resources and the app registration, as described in Prepare Azure Resources for Kubernetes Edge.

  3. Allow the outbound connections listed in Kubernetes Edge Network Access Requirements.

  4. Register the edge instance in the Ataccama Cloud Portal.

  5. Install the Kubernetes edge with Helm.

Once the edge shows as connected in the Cloud Portal, select it in your data source connections to process data on it. See Configure data sources to use the edge.

Was this page useful?