>>The second link gives the best description of what is going on.
>
>I am trying to understand this concept. Right now, I open a connection when the Web site hit starts. During the hit, it can process from 1 to x number of SQL requests. Then, at the end of the hit, I close the connection.
>
>So, basically, a connection does not remain opened after the hit. Using a connection pool seems to mean that we add a concept of reusing active database connections instead of creating a new connection with every request. Well, if I close them at each hit, how can it be reused?

In this context 'Close' doesn't mean it is actually closed - just returned to the pool. Likewise 'Open' doesn't neccessarily mean a connection is opened- most often an existing connection from the pool is used. So the overhead for opening and closing after each query is not that large.