For RHODS disconnected, we want to support the same model that's supported for OpenShift disconnected. This is covered in detail here. The rationale is that RHODS is an add-on to OCP so customers will expect RHODS & OCP to follow similar models for disconnected support. A few key points:

1. RHODS self-managed should be fully functional when used in an environment that is not connected to the internet. The experience and all functionality should be identical when operating in a disconnected or connected environment. 

2. For installation & updates, RHODS should support the same methods as OCP for disconnected/air-gapped installs.

• Customer sets up an internal/mirrored registry in their data center.
• While connected, the required images are pulled from an external registry to the internal registry.
• The internal registry can be connected and disconnected. If a customer wants to get updates, they can connect the internal registry to get the updates, then disconnect.
• The installation process should be similar except that it points to the internal mirrored registry and downloads from that registry.

Resources:

• Install overview: https://www.youtube.com/watch?v=iUkyElVdK7I
• Upgrade overview: https://www.youtube.com/watch?v=_o-ZH6qXvtc 

Clone for epics and close what is not needed

Things to consider and add to the below branches as needed:

• Implement component testing framework (for new components)
• Add tutorials and documentation to ODH website
• Implement Monitoring
• All new features/components must have an upgrade strategy:
• Base language version (e.g. golang versions will be aligned with openshift core components or only use LTS).
• Dependencies

• Write standard operating procedures for new feature for SRE

• Static analysis for new repos
  • For static analysis, we are working with QE on what framework is best. 
• Blog post about new feature and component
• Record new feature and post to youtube
• What open source projects could come out of this work
• Does it impact any Tier 2 component integration that we need to make the owners of those components aware of?
• For new components:
• Update/create a standard operating procedures document for MT-SRE.  Example SOP .  Check with SRE team and AI services managers.
• Update/create training material.  Dave Mulford is a good contact for this.