Kloudfuse Install Prequisites

Prerequisites

Before you begin, ensure the following requirements are in place:

Requirement Details

Kubernetes cluster

A cluster running on GCP (GKE), AWS (EKS), or Azure (AKS). The cluster can be dedicated to Kloudfuse or shared with other application workloads.

Node resources

At least 16 vCPUs and 128 GB memory available across the cluster. For production deployments, see Cluster Configuration to size and isolate node pools by workload type.

Persistent Volume Claims

The cluster must support dynamic PVC provisioning. Kloudfuse uses persistent volumes for its storage components.

Helm

Version 3.10.0 or later. Used to install and upgrade the Kloudfuse stack.

kubectl

Configured to communicate with your target cluster.

Installation steps

Follow these steps in order for a complete Kloudfuse deployment:

Step Description

Install Using Helm

Deploy the Kloudfuse Helm chart with your custom configuration values.

Ingress Configuration

Configure cloud provider-specific Helm values to expose the Kloudfuse UI and API endpoints.

Local Authentication

Add initial user accounts for platform access.

SSO Authentication

Optional: configure Single Sign-On to integrate with your identity provider (Google, Okta, JumpCloud, or SAML).

Pinot Object Storage

Configure an external object store (S3, GCS, or Azure Blob Storage) for long-term data retention in Pinot.

Uninstall Kloudfuse

Remove the Kloudfuse stack when needed.

Advanced configuration

For production deployments, review these additional topics after completing the initial installation:

Topic Description

Cross-AZ Recovery

Configure backup, restore, and disaster recovery procedures to protect your Kloudfuse data.

HA Multi-AZ Setup

High Availability(HA) setup for Kloudfuse, you will deploy the cluster across multiple availability zones (multi-AZ)

Kloudfuse Authentication

Configure SSO, SAML, ingestion authentication, and login security policies.

Cloud-Provider Database

Optional: configure Kloudfuse to use a cloud provider-managed database instead of the bundled database.

Cluster Configuration

Isolate ingestion, query, and control plane workloads across dedicated node pools. Configure Pinot stream isolation for large-scale deployments.