Design: Abstracting JDBC and datasource to SQL/JTable

My back-end calls for a dynamic datasource and topology, and my front-end calls for a bunch of app coding that queries the RDBMS. My idea is to walk away from the JDBC API altogether, as it turns out, for any API the app coding couples itself to.

Here's the design idea... On queries, the app code passes in SQL, and receives JTable as a result. On updates, the app code passes in SQL (even batches of it), and receives success/failure code (plus error message) as a result.

Would you couple all of your app code to class JTable? Would there be a better class for generic, decoupled query result handling? (If I ever need metadata, I can subclass/encapsulate that in with the result) Are there examples out there of this already being done? I do not need more than the above for this application. I don't have extravagant RDBMS needs. TIA - Mark