The reason that I am not designing the system with direct dbase access is that the system will need to be used in offline situations where the user does a syncronization and then goes out on site with a laptop. The user can make changes offline and then, we he/she gets back online, can syncronize. This was the other reason I was looking at datasets due to their ability to dump to xml and schema (and import from same) which I was using for offline cache. I may look into have a table on the dbase that records all changes to any other of the dbase tables hence, rows updated since the last update could be downloaded easily. This might be a way to handle it.