Just the Facts

• ADO.NET data container classes such as DataSet and DataTable have no notion of the provider that served them any data. They are serializable, feature-rich classes and offer a bunch of database-like functions, such as referential integrity, optimistic locking, constraints, indexing, and filtering.

• The data adapter is a command-like object that performs data access, grabs all the data, and packs it into a data container. Like commands and data readers, data adapters are specific to each data provider.

• The Fill method of the data adapter maps the source result sets onto in-memory tables and fills the tables with the data fetched from the physical data source.

• In ADO.NET 2.0, you can better control the way data is loaded into the various data tables during a fill operation. You can choose, for example, to override current or underlying values.

• DataSet is great at encapsulating tables of data and relationships and moving it between the tiers of an application in a loosely coupled manner.

• ADO.NET batch update consists of a series of commands that the data adapter submits sequentially to the database. Batch update is triggered by a single instruction but doesn't necessarily equate to a series of queries submitted in a single command.

• Batch update is particularly effective in environments with a low degree of data contention because it allows for disconnected architectures, higher scalability, and considerably simpler code.

• In ADO.NET 2.0, DataSet can be serialized in a true binary format, which gets you a much more compact output.

• In ADO.NET 2.0, DataTable implements the IXmlSerializable interface and provides public methods to load and save its contents from and to XML streams. The implementation of the interface also enables DataTable to be used as a parameter and return value from .NET Web service methods.