FAQ
Quick answers to the questions platform teams ask most when standing up and running Nexa in their own cloud. Each answer links to the page with the full detail. If your question isn’t here, see Getting Support.
Deployment
Section titled “Deployment”Where does Nexa run — DataReadyAI’s cloud or ours?
Section titled “Where does Nexa run — DataReadyAI’s cloud or ours?”Yours. Nexa is customer-managed: DataReadyAI ships container images and infrastructure-as-code, and your platform team installs and operates it in your own AWS or Azure account against your Snowflake or Databricks warehouse. DataReadyAI supports you but has no access to your environment. See Deployment model.
Which clouds and data platforms are supported?
Section titled “Which clouds and data platforms are supported?”Cloud: AWS or Azure. Data platform: Snowflake or Databricks. You pick one of each per deployment; the platform-variant differences are handled by the relevant API service (nexa-databricks-api or nexa-snowflake-api). See Reference architecture.
How are services deployed and updated?
Section titled “How are services deployed and updated?”Through GitOps: ArgoCD syncs Kustomize overlays from the nexa-deployments repository to your Kubernetes cluster (EKS/AKS). Non-production environments auto-update to new image tags; production is a deliberate, manual promotion. See the Upgrade Guide and Installation.
What do I need before I can install?
Section titled “What do I need before I can install?”A Kubernetes cluster and cloud account (provisioned by the IaC), a container registry with the Nexa images, a data-platform connection (Snowflake or Databricks), and DNS/TLS for the web endpoint. The full checklist is in Prerequisites.
Security
Section titled “Security”Does DataReadyAI have access to our data or environment?
Section titled “Does DataReadyAI have access to our data or environment?”No. Everything runs in your account; DataReadyAI cannot reach your cluster, database, secrets, or warehouse. Support is collaborative — you apply the fixes because only you have access. See Deployment model.
Where are secrets and credentials stored?
Section titled “Where are secrets and credentials stored?”In your cloud secret store (AWS Secrets Manager / Azure Key Vault), surfaced to pods through the Secrets Store CSI driver via a SecretProviderClass per service — not baked into images or committed to Git. See Configuration Reference and Security & network.
How is data isolated across business units?
Section titled “How is data isolated across business units?”Isolation is physical, not row-level: sensitive scopes get their own indexes/resources rather than relying on row filters on a shared object. Governance features (business glossary, PII tagging, change review) sit on top. See Security & network.
Where do I report a security vulnerability?
Section titled “Where do I report a security vulnerability?”To the security contact, privately — not in a public issue or general support channel. See the security row in Getting Support.
Operations
Section titled “Operations”How do I monitor health and know a deployment is healthy?
Section titled “How do I monitor health and know a deployment is healthy?”Each service exposes liveness/readiness probes and health endpoints; ArgoCD reports sync/health status, and the cluster reports pod state. Key metrics, thresholds, and monitoring setup are in the Operations Guide.
What gets backed up, and how do I restore?
Section titled “What gets backed up, and how do I restore?”The application Postgres database holds all Nexa metadata (glossary, pipeline and agent definitions, cost history) and is the primary thing to back up; your warehouse data lives in Snowflake/Databricks and is managed there. Backup and restore procedures are in the Operations Guide.
Something’s broken — where do I start?
Section titled “Something’s broken — where do I start?”Start with the symptom → cause → fix table in Troubleshooting & Diagnostics, which also has the commands to collect a diagnostic bundle. If it started right after an upgrade, check the Release notes for a known issue.
How do I roll back a bad upgrade?
Section titled “How do I roll back a bad upgrade?”Revert the image-tag change in Git and re-sync ArgoCD (preferred), or use argocd app rollback / kubectl rollout undo for a faster revert. Note that forward-only database migrations are not undone by an image rollback — restore from backup if a migration ran. See the Upgrade Guide.
Billing and licensing
Section titled “Billing and licensing”Who pays for the cloud resources?
Section titled “Who pays for the cloud resources?”You do — all compute, storage, database, and data-platform usage is in your account and billed by your cloud and warehouse providers directly. DataReadyAI does not resell or meter your infrastructure. To stop that billing when decommissioning, follow the Uninstall / Teardown steps.
How is the Nexa software licensed?
Section titled “How is the Nexa software licensed?”Under your DataReadyAI agreement, separate from cloud infrastructure cost. Licensing specifics (entitlements, environments covered, support tier) are defined in that agreement; contact your delivery representative via Getting Support for questions.
How do I track and control platform spend?
Section titled “How do I track and control platform spend?”Nexa includes cost-tracking dashboards for warehouse and compute spend under Governance, and you can scale services up or down per environment. Scaling signals and cost-control guidance are in the Operations Guide.