Skip to main content

Documentation Index

Fetch the complete documentation index at: https://arize-ax.mintlify.dev/docs/llms.txt

Use this file to discover all available pages before exploring further.

​
Overview

Use this page when you already have a GKE cluster and will create or align supporting resources (buckets, service accounts, node pools) to match Arize’s expectations.

​
Storage buckets

Create two Google Cloud Storage buckets: one for gazette data and one for druid data. Their names are referenced in values.yaml as gazetteBucket and druidBucket.

​
Node pools

Use at least two node pools:
  • Base pool — general workloads and ingestion. Label nodes with arize=true and arize-base=true.
  • Druid pool — ArizeDB historical workloads. Label nodes with arize=true and druid-historical=true.
If you cannot use separate pools, you can deploy on a shared pool by setting historicalNodePoolEnabled: false in values.yaml (see Prerequisites for sizing context).

​
Storage classes

Prefer premium-rwo and standard-rwo where available; these are used by default in the GCP distribution.

​
Service account and permissions

Attach a GCP service account with permissions that allow Arize to use GCS, BigQuery jobs, Artifact Registry pulls, and Vertex AI predictions as required by your deployment. Typical permission families include:
  • bigquery.jobs.create
  • storage.objects.create, storage.objects.delete, storage.objects.get, storage.objects.list
  • artifactregistry.repositories.downloadArtifacts
  • aiplatform.endpoints.predict
Exact bindings should follow your organization’s least-privilege standards. If you use Workload Identity, grant the Kubernetes service accounts the ability to impersonate the GCP service account (roles/iam.workloadIdentityUser) for:
  • arize/arize
  • arize-operator/arize-operator
  • arize-spark/spark

​
JSON key (no Workload Identity)

If you do not use Workload Identity, provide the service account JSON key via Helm values (base64-encoded) as documented in the detailed walkthrough (GCP).

​
Namespaces

The Helm chart can create or use existing namespaces: arize, arize-operator, and arize-spark.

​
Optional image registry

A GCR or Artifact Registry is optional when pulling from Arize’s central registry. If you use a private registry, set pushRegistry and pullRegistry in values.yaml.

​
Next steps