Why is a Merkle proof difficult to implement in a relational DB?

The only required data fields in each row are the parent row ID and a hash, no?