I’m not close enough to any of the ledger implementation to give you specific details, but I can tell you one thing they all have in common: we consider the on-ledger storage format to be an internal implementation detail, and users should only ever interact with DAML data through the DAML API.
As explained in this thread, for the case of Fabric, we’re not actively trying to make it impossible for people to get access to the data (at least for ledgers on which the data is not encrypted), but it is unsupported, subject to change, and all the other bad things that come with relying on implementation details.