Client/server: browsing the web

Figure fig:websession shows a typical session for downloading a web page containing some (in this case two) embedded images. Some details concerning the establishment of TCP connections are left out. Some details concerning HTTP (Hypertext Transfer Protocol, [HTT99]) are mentioned. The following steps occur:

Illustration 2.1: Client requesting a document from a web server

1. The client (in this case the web browser) tries to connect to the web server, usually using port number 80.
2. The server accepts the connection.
3. The client requests a web page (in this case index.html). This is performed by using the GET command of HTTP.
4. The server responds by sending the requested web page. The server also sends (as a part of the reply) an indication that the operation was successful.
5. The client parses the response and detects that there are some embedded images on this web page.
6. The client issues requests for these images to the server.
7. The server responds by sending the images to the client.
8. The client closes the connection to the server.

The older version of HTTP, version 1.0 ([HTT96]), stated that the client closes the connection between each request and hence needed to reestablish the connection with the server to request the embedded images. This made the protocol very simple, but somewhat inefficient. The later version of HTTP, version 1.1 ([HTT99]), allowed clients to maintain a session with the server, making it possible to request the embedded images using the same connection.