If you are wondering why HTTP works this way, it has to do with the unpredictability of users.
While a user is looking at my main Welcome page they are quite likely to follow a link to my RelationaldbDesign link.
A user is just as likely to type in a URL for CNN, or follow a bookmark to a search engine. If a Web server keeps a connection open, waiting for another request, that request might never come.
The browsers would have to send You can hang up, my user has gone elsewhere messages to the server.
Servers would be completely consumed by maintaining unused connections.
This is wasteful in a world where users do not stay very long at each Web site.