What is Connection pooling?

The connection represents an open and unique link to a data source. In a distributed system, this often involves a network connection. Depending on the underlying data source, the programming interface of the various connection objects may differ quite a bit. A connection object is specific to a particular type of data source, such as SQL Server and Oracle. Connection objects can't be used interchangeably across different data sources, but all share a common set of methods and properties grouped in the IDbConnection interface.
In ADO.NET, connection objects are implemented within data providers as sealed classes (that is, they are not further inheritable). This means that the behavior of a connection class can never be modified or overridden, just configured through properties and attributes. In ADO.NET, all connection classes support connection pooling, although each class may implement it differently. Connection pooling is implicit, meaning that you don't need to enable it because the provider manages this automatically.
ADO.NET pools connections with the same connection or configuration (connection string). It can maintain more than one pool (actually, one for each configuration). An interesting note: Connection pooling is utilized (by default) unless otherwise specified. If you close and dispose of all connections, then there will be no pool (since there are no available connections).
While leaving database connections continuously open can be troublesome, it can be advantageous for applications that are in constant communication with a database by negating the need to re-open connections. Some database administrators may frown on the practice since multiple connections (not all of which may be useful) to the database are open. Using connection pooling depends upon available server resources and application requirements (i.e., does it really need it).
Using connection pooling
Connection pooling is enabled by default. You may override the default behavior with the pooling setting in the connection string. The following SQL Server connection string does not utilize connection pooling:
Data Source=TestServer;Initial Catalog=Northwind;
User ID=Chester;Password=Tester;Pooling=False;
You can use the same approach with other .NET Data Providers. You may enable it by setting it to True (or eliminating the Pooling variable to use the default behavior). In addition, the default size of the connection pool is 100, but you may override this as well with connection string variables. You may use the following variables to control the minimum and maximum size of the pool as well as transaction support:

� Max Pool Size: The maximum number of connections allowed in the pool. The default value is 100.
� Min Pool Size: The minimum number of connections allowed in the pool. The default value is zero.
� Enlist: Signals whether the pooler automatically enlists the connection in the creation thread's current transaction context. The default value is true.

The following SQL Server connection string uses connection pooling with a minimum size of five and a maximum size of 100:

Data Source=TestServer;Initial Catalog=Northwind;
User ID=Chester;Password=Tester;Max Pool Size=50;
Min Pool Size=5;Pooling=True;