TODO

# Hermes Interaction
    # Extract Hardware Metadata
        1) Use Jarvis to collect the cluster devices (Resource graph)
        2) Use Hermes to get the current status of all the tiers in the hierarchy. How?\
            RPC std::vector<TargetInfo> LocalGetTargetInfo()
            metadata_manager.h
            HERMES->mdm_->LocalGetTargetInfo
    # Extract software metadata
        3) Use Hermes to get the current location/size/... of all the blobs and buckets on the system.
    # Leverage a customizable Hermes
        4) Custom DPE
        5) Custom PreFetching

# Derived Variables
    # Aid for ORNL
        # What are this?
        # Get an example?
        # Can we get a Query
    # Hermes
        # Tratis. How to define them, how to apply them
        # Leverage the custom DPE and Pretching to handle them differently
    # Application level
        # How should applications define them?
        # Should we be transparent or requires changes to the code?