Installing the PG AI Hybrid Manager

PG AI Hybrid Manager (HM) provides orchestration, observability, and lifecycle operations for Postgres, analytics, and AI workloads across cloud and on-premises deployments.

This section provides installation guidance for users deploying Hybrid Manager on their own infrastructure, across supported Kubernetes platforms. If you're using a turnkey hardware system like Sovereign AI and Data Factory, Hybrid Manager is preinstalled and requires no setup.

Before you install

Hybrid Manager requires a Kubernetes environment and access to platform-specific APIs for managing storage, networking, and node pools.

You can install Hybrid Manager in:

• Self-managed Kubernetes (on-premises or cloud)
• RedHat OpenShift clusters
• Cloud-managed Kubernetes (EKS, GKE)

Make sure to:

• Review system resource and storage requirements
• Choose a supported Postgres distribution (Advanced Server, Extended Server, PostgreSQL)
• Define access and RBAC requirements for your organization

Planning

Kubernetes (Generic)

Use this guide to prepare a custom Kubernetes cluster for Hybrid Manager.

• Planning your Kubernetes cluster for HM installation

This includes sizing recommendations, storage class expectations, and node group setup.

Installation guides by platform

Choose your target platform to follow the appropriate installation procedure.

RedHat OpenShift (RHOS)

Hybrid Manager is certified for deployment on OpenShift clusters.

• Install HM on RHOS

Amazon Web Services (EKS)

Install Hybrid Manager on Amazon Elastic Kubernetes Service (EKS).

• Install HM on EKS

Google Cloud Platform (GKE)

Install Hybrid Manager in a GKE-managed Kubernetes cluster.

• Install HM on GKE

Considering a preinstalled hardware solution?

If you're looking for a fully managed deployment of Hybrid Manager and its supporting infrastructure, consider the Sovereign AI and Data Factory — a turnkey hardware solution delivered by EDB.

This system includes:

• Preinstalled, end-to-end support of Hybrid Manager and Redhat Openshift
• Full lifecycle management of hardware and software
• Built-in support for Postgres, vector workloads, and GPU-powered inference

Learn more at the Engineered Systems overview.