At high level, this is what occurs. Beyond this, I would suggest diving into the PTC University courses as fast as possible to backfill your Windchill knowledge.
CAD files are saved to a workspace which is a structured folder on the user's system (meta data and obfuscated files). There is a "workspace" object on the server side. Before a check in starts, CAD files are uploaded and placed in a CAD Document object in that workspace or the users personal area. No one else but admins can see them. The files would normally be stored to the default cache vault as a temporary landing location. During the check in step, the object status on the server is updated to be visible to others and moved into a Library or Product area "folder". This files that are associated with this CAD Document are then moved from the default cache vault to the main vault depending on the vaulting rules of the Product or Library. The Database keeps track of where those files are.
Depending on your WVS rules, on check in, a visualization job is kicked off to a CAD Worker. They typically operate like normal users. They open Creo, download the files to a workspace and publish content for Creo View. That data is packed up and sent back to the server. More visualization objects are created, linked to the CAD Document and the published content, guess what, is stored to the vaults.