Objects consist of two parts when running - process and state (called ObjectPersistentRepresentation)
Running object processes must be able to directly access their OPR in order to save and restore their state
Our current OPR implementation uses local files (Unix or NT) as the backing store and identifies OPRs by path name
Consequently, objects must have direct access to their OPR directory using the name the Vault knows them by.
To facilitate scheduling we track host/vault compatibility
- If a process starting on Host A can directly access the OPRs stored by Vault B, then Host A and Vault B are compatible.