ADO.NET is the data access model for .NET-based applications. It can be used to access relational database systems such as SQL Server 2000, Oracle, and many other data sources for which there is an OLE DB or ODBC provider. To a certain extent, ADO.NET represents the latest evolution of ADO technology. However, ADO.NET introduces some major changes and innovations that are aimed at the loosely coupled-and inherently disconnected-nature of Web applications. For a comparison of ADO and ADO.NET.
One of the key changes that ADO.NET introduces is the replacement of the ADO Recordset object with a combination of the DataTable, DataSet, DataAdapter, and DataReader objects. A DataTable represents a collection of rows from a single table, and in this respect is similar to the Recordset. A DataSet represents a collection of DataTable objects, together with the relationships and constraints that bind the various tables together. In effect, the DataSet is an in-memory relational structure with built-in XML support.
One of the key characteristics of the DataSet is that it has no knowledge of the underlying data source that might have been used to populate it. It is a disconnected, stand-alone entity used to represent a collection of data, and it can be passed from component to component through the various layers of a multitier application. It can also be serialized as an XML data stream, which makes it ideally suited for data transfer between heterogeneous platforms. ADO.NET uses the DataAdapter object to channel data to and from the DataSet and the underlying data source. The DataAdapter object also provides enhanced batch update features previously associated with the Recordset.
This Figure shows the full DataSet object model.