Feature Overview (aka. Goal Summary)  

Design a solution / Proof of concept for how to optimize disk space.

Goals (aka. expected user outcomes)

The new implementation of `oc-mirror v2` is a significant amount of disk space compared to `v1`. The goal of this issue to provide a better design

Background

One of the big complains in v1 was to start the workflow always from scratch when something failed during the process. We had customers losing hours of work because something failed at the end of the mirroring and they needed to start from scratch. 

This is how the cache was born in v2, so they can continue the mirroring from where it stopped. The downside is that it required additional disk space.

https://github.com/openshift/oc-mirror/issues/1123

NAPS feedback (300GB is required).

Requirements (aka. Acceptance Criteria):

• This is a spike.
• What is the current amount of disk space used to mirror a typical OCP deployment? (full catalog is 4TB and 15GB each release, but I assume customers do not mirror the full catalog).
• What was the disk space used with v1
• Based on the two, review the current implementation and consider solutions to reduce the disk space

Also consider related RFEs in the design: RFE-7904,  RFE-7059, RFE-6823 .

Documentation Considerations

N/A