Prerequisites for Installing Gaia AI Engine

Deployment Model: Self-Managed

29 January 2026

This section describes the requirements and prerequisites for installing Gaia AI Engine.

Before proceeding, ensure the following components are set up and configured.

Install and Set Up Helios Self-Managed

Ensure that you have installed and set up Helios Self-Managed on your physical appliance.

For more information, see Set Up Helios on Your Physical Appliance.

Install the Gaia Service

Ensure that you download and install Gaia Service version 1.3.1.

For more information, see Manage Services.

Set Up the Red Hat OpenShift Cluster

Deploy and configure a Red Hat OpenShift cluster.
Install persistent volume provisioners.
Set up a load balancer provider. For example, you can use MetalLB.

For more information, refer to the Gaia Self-Managed Deployment Guide.

Configure Networking

Reserve an IP address for use by the load balancer provider configured above.
Ensure that the Cohesity cluster can reach this IP address.
Ensure that all Cohesity clusters connected to Helios Self-Managed that will interact with the Gaia AI Engine can reach this IP address.
Create a DNS record that maps to this IP address.
When you register the Cohesity cluster in Helios Self-Managed, it operates within a designated subnet. If the Gaia AI Engine is deployed on a Kubernetes cluster located in a different subnet, you must ensure that there is network connectivity between the Cohesity cluster and the Kubernetes cluster hosting Gaia. This connectivity is essential to enable seamless communication between the clusters.
Ensure that there is network connectivity between the Cohesity cluster and the Gaia AI Engine, and between the Gaia AI Engine and the S3 endpoint.
Ensure that your network firewall allows outbound traffic from the Cohesity cluster to the Kubernetes cluster where Gaia AI Engine is deployed.
Ensure that the DNS name assigned to the Gaia AI Engine resolves to a reachable IP address from all required networks and the associated port is allowed in all relevant firewalls.

Install and Configure NVIDIA GPU Drivers

Deploy the Node Feature Discovery (NFD) Operator. For more information, see Installing the Node Feature Discovery Operator on OpenShift.
Deploy the NVIDIA GPU Operator. If the nodes are connected to the internet, the operator automatically installs NVIDIA drivers. Otherwise, manually install the drivers on each node. For more information, see Installing the NVIDIA GPU Operator.

Set Up Container Registry

Choose a Docker container registry.
Ensure that the Kubernetes cluster where the Gaia AI Engine is deployed can reach the container registry.

Configure S3-Compatible Storage

When indexing a dataset, Gaia generates embeddings and text. In a Self-Managed deployment, you must configure S3-compatible storage to store this data. To set up the storage, gather the following details:

S3 service hostname and port number
Bucket name
Access key
Secret access key
Determine and download the CA certificate for connecting to this S3 endpoint.

You can use the following command to identify the certificate used by the S3 service:
```
openssl s_client -showcerts -connect <hostname>:<port>
```
Replace <hostname> and <port> with the appropriate values for your S3 service.
Verify S3 connectivity from the Gaia AI Engine to the S3 endpoint.

Cohesity recommends creating an S3 View using Cohesity SmartFiles to store the embeddings and text. When creating the S3 View, select Backup Target SSD as the QoS policy and enable the Pin Views to SSD option. For more about these options, see Performance.
For more information about creating an S3 View, see Create an S3 View.

Gaia AI Engine Package

Download the Gaia AI Engine tar file from the Cohesity Downloads portal.

Insights Documentation