Deploying VioGraph

VioGraph is Kubernetes-native. The full platform runs inside your cluster, under your credentials — we never see your data, your queries, or your warehouse credentials.

Supported platforms

• GKE (Google Kubernetes Engine) — fully supported
• EKS (Amazon Elastic Kubernetes Service) — fully supported
• AKS (Azure Kubernetes Service) — fully supported
• OpenShift — supported (use openshift-router ingress)
• k3s (on-prem / edge) — supported for small deployments

Minimum requirements

Component	CPU	RAM	Storage
Control plane	500m	1Gi	10Gi (for Postgres metadata)
Export worker	500m	2Gi	ephemeral
Graph wrapper (per instance)	1	4Gi+ (depends on graph size)	ephemeral

Wrappers are ephemeral — they spin up per query batch and shut down when idle. The control plane and export worker are always-on but small.

Install

helm repo add viograph https://helm.viograph.io
helm install viograph viograph/viograph \
  --namespace viograph --create-namespace \
  --set ingress.host=graph.yourcompany.com

Configuration

Key values to customise in values.yaml:

• ingress.host — public hostname
• ingress.tls.secretName — TLS cert secret (use cert-manager in prod)
• auth.oauth2Proxy.enabled — enable SSO (Okta / Azure AD / Keycloak)
• database.external.enabled — use an external Postgres instead of the in-cluster one
• wrappers.resources.limits — graph pod size limits

See the full values reference for every option.

Data residency

All data — warehouse credentials, graph data, query history — stays inside your cluster. VioGraph reads from your warehouse using your connection and writes ephemeral results to a local PVC that is wiped when the graph is terminated.

Nothing leaves your VPC. Apache 2.0 licensed so your security team can audit every line before deploying.