Image and Model Library Explained

Image and Model Library Explained#

Image and Model Library Explained#

The Image and Model Library is a core capability of the Hybrid Control Plane (HCP). It provides a unified, governed repository for managing container images across your entire HCP environment.

This includes:

  • Database images → for PostgreSQL cluster provisioning.

  • AI model images → for Model Serving (via AI Factory Model Library).

Why a single library?#

Using a unified Image and Model Library ensures:

  • Governance → all images (DB or AI model) pass through the same approval flow.

  • Consistency → the same agent-driven sync pulls images from container registries.

  • Security → private registry integration applies equally to all image types.

  • Flexibility → both database and AI teams can manage images within a common framework.

What is the difference between Image Library and Model Library?#

Capability

Image and Model Library (Hub)

Model Library (AI Factory)

Scope

All container images used in HCP

Subset: AI model images used for Model Serving

Primary Users

DB admins, platform teams, AI teams

AI Factory users, AI engineers

Backed by Container Registry?

Yes

Yes (via Image Library)

Supports Private Registries?

Yes

Yes (via Image Library)

Source of Truth

Container Registry → Image Library

Filtered view of Image Library

Usage

Postgres cluster provisioning, AI model serving

AI model serving only

UI Location

Image Library (HCP Portal)

Model Library (AI Factory UI)

Summary: → The Model Library is powered by Image Library. → All images pass through Image Library first → Model Library presents a user-friendly view of AI model images.

Architecture flow#

Container Registry → Image and Model Library →
→ Postgres Provisioning / AI Model Serving
→ Model Library UI → Model Serving

  • Postgres provisioning consumes database images from Image Library.

  • AI Factory Model Library consumes model images → curated from Image Library.

Key Concepts Concept Meaning Repository A collection of images (ex: postgresql, meta/llama-3.3-49b) Tag A version of an image (ex: 17.4-2503310734, 1.8.5, latest) HCP Tags Organizational tags applied to image repos in HCP Repository Rules Define which repos/tags to sync from connected registries Private Registry Any customer-provided container registry integrated into HCP Beacon Agent Agent that pulls image metadata into Image Library

Use Cases Database images Provision new Postgres clusters using curated database images.

Support for:

PostgreSQL Community.

EDB Postgres Advanced Server (EPAS).

EDB Postgres Extended Server (PGE).

Manage extensions via OCI annotations.

AI model images Deploy AI models using AI Factory → Model Serving (powered by KServe).

Support for:

NVIDIA NIM model images.

Future: customer-provided model images.

Model Library surfaces a curated view of AI model images from Image Library.

Governance flow Connect container registries → private or public.

Define Repository Rules → which repos/tags to sync.

Beacon Agent pulls images → Image Library populated.

Image Library available to:

Postgres Provisioning flows.

Model Library (for AI model serving).

Model Library filters Image Library → only AI models shown.

Benefits

  • Single governance model for all images.

  • No bypass of security or compliance policies.

  • Private registry integration → works equally for database and AI model images.

  • Teams can manage their own repositories but operate within platform-wide visibility.

Related Explained Pages Container Registry Integration Explained

AI Factory Model Serving Explained (Model Library)

Related How-Tos

  • Integrate Private Registry

  • Define Repository Rules

  • Find and Use Images

  • Deploy AI Models

  • Provision a Postgres Cluster with Custom Image

Contents