Container Registry Integration

The Image and Model Library in Hybrid Control Plane (HCP) is powered by container registries. Registries provide the actual image content for:

• Database images (PostgreSQL, EPAS, PGE)
• AI model images (NVIDIA NIM, future custom models)

Understanding how container registry integration works is key to managing and extending your Image Library.

How does it work?

HCP relies on the Beacon Agent to pull image metadata from connected registries.

The Beacon Agent uses Repository Rules to know:

• Which registries to connect to
• Which repositories to scan
• Which image tags to import

→ The Image Library UI in HCP is a reflection of what the agent has pulled.

Key Concepts

Default registries

In ephemeral HCP environments:

• Default registry is upmdev.azurecr.io (Azure ACR).
• Repositories:
• postgresql → PostgreSQL community images.
• edb-postgres-advanced → EPAS images.
• edb-postgres-extended → PGE images.

In customer environments:

• Customer provides a private registry.
• Image Library is configured to pull from that registry.

Repository Rules

Repository Rules drive what Beacon Agent discovers:

• You can add custom repositories via Image Library UI → "Manage Repositories."
• You can edit the Beacon Agent config for advanced control.

Example agent config section:

dispatcher:
  capabilities:
    provisioning:
      images:
        discovery:
          registry:
            postgres_repositories:
              - name: postgresql
                description: Community PostgreSQL
              - name: edb-postgres-advanced
                description: EPAS
              - name: edb-postgres-extended
                description: PGE
              - name: postgresql-torsten
                description: Custom PG Build with AIDB

This tells agent which repositories to pull and display.

→ You can edit this config to add more repositories (advanced).

Private Registry Integration Flow Add repository via Image Library UI → simplest flow for most users.

Agent pulls image list periodically.

Image tags and metadata appear in Image Library.

You can select these images for:

Database provisioning

AI model serving (via Model Library in AI Factory).

Relationship to AI Factory Model Library Model Library in AI Factory is powered by Image Library.

All AI model images flow through this same registry → agent pull → Image Library → available for Model Serving.

This ensures consistent governance of all model images.

→ See: AI Factory Model Serving Explained Summary

• Registries provide the source of truth for images.
• Repository Rules tell HCP which repos/tags to scan.
• Beacon Agent pulls the images and updates Image Library UI.
• Private registries can be fully integrated and governed.
• AI Factory Model Library leverages this same system.

Related How-Tos Integrate Private Registry

Define Repository Rules

Related Explained Pages Image and Model Library Explained

AI Factory Model Serving Explained