A bit general this...

I've recently started working on a reasonable sized system. The only access
I have to the database files is through a limited ODBC driver. I have started
working on a client / server implementation where a request is sent to be
server for a group of records, these are then put end to end into a string,
passed back and fed into appropriate class structures for the client to use.

I still have to write code to enable data to be written back and I was wondering
if I was approching the project in the right way, is it actualy that much
more efficient to write in this way, or would it have been easier to put
the database code in the client ?

Any thoughts would be much appreciated

TIA

John Greenwood