Skip to content
For Self-Managed AI Platform users running v11.1, see the on-premise platform documentation

Amazon Elastic Kubernetes Service (EKS)

This section lists the platform-specific requirements and outlines the steps necessary to install the DataRobot platform on an Amazon Elastic Kubernetes Service (EKS) cluster.

Terraform infrastructure module

To minimize manual operations and allow for simplified configuration, DataRobot provides a dedicated Terraform module for AWS infrastructure.

This module automates the setup of critical components, including:

  • Networking: Creating a Virtual Private Cloud (VPC), subnets, and NAT Gateways.
  • Kubernetes: Provisioning an Amazon Elastic Kubernetes Service (EKS) cluster.
  • Storage: Configuring S3 buckets for persistent file storage.
  • Registry: Creating Amazon Elastic Container Registry (ECR) repositories for custom images.
  • Security & DNS: Setting up necessary IAM roles, Route 53 DNS zones, and KMS encryption keys.
  • Networking add-ons (via Helm charts):
    • ingress-nginx for managing external access to the Kubernetes cluster.
    • cert-manager for automated certificate management (e.g., Let's Encrypt).
    • external-dns for automatically creating DNS records for Kubernetes services.
    • aws-load-balancer-controller for managing AWS load balancers.
    • cluster-autoscaler for automatically adjusting the size of the EKS node groups based on the resource demands of running pods.
    • descheduler for evicting and rescheduling pods to ensure they are balanced across nodes according to current cluster policies and resource availability.
    • ebs-csi-driver for allowing the EKS cluster to manage the lifecycle of Amazon EBS volumes as persistent storage for Kubernetes pods.
    • nvidia-device-plugin for exposing GPU resources to the cluster, allowing Kubernetes to schedule and run GPU-accelerated workloads.
    • metrics-server for collecting resource metrics like CPU and memory usage from nodes and pods to enable features like the Horizontal Pod Autoscaler.

It is intended to simplify and standardize the infrastructure layer required before installing the DataRobot platform via Helm charts.

Platform-specific requirements

In addition to the generic infrastructure requirements, you need the following platform-specific requirements for installing DataRobot in the AWS cloud:

Preparation

Before you can install the DataRobot platform, you must pull the DataRobot container images and push them to your private OCI registry:

設定

Prepare your values_dr.yaml file with all the necessary configurations for your environment:

インストール