Wow Bonnie, you have *really* summarized n-tier architecture extremely well in this thread (not that I'm surprised *s*). Perhaps you'd like to post a similar paragraph for how data updates would be handled. It's really about the same of course as far as the layers and the fact that they are passing around a dataset or datarecord, but a quick summary (like below) of where validation of rules and anything else pertaining to inserts, updates and deletes (such as concurrency) would really round out this thread, don't you think? *g*

Thanks for these and your other fine posts here, Bonnie.

Kelly

>The UI needs Customer data. It instantiates the MyCustomerBiz class and calls it's GetCustomer() method. The GetCustomer() method in the MyCustomerBiz class instantiates the MyCustomerDataAccess class and calls it's GetCustomer() method. The GetCustomer() method in the MyCustomerDataAccess class takes care of all the DataAdapters stuff and Fills a CustomerDataSet. It passes the CustomerDataSet back to the MyCustomerBiz class. If there's any additional Biz logic that needs to be done to the dataset, it's done here. Then the MyCustomerBiz class passes the CustomerDataSet back to the UI.
>
>The advantage of having separate MyDataSet project is that you can define the CustomerDataSet separately in it's own class. Then each of the three layers can use the same DataSet class.