If two computers on the Internet exchange data, certain conventions
must be observed, so that the computer receiving the data can process
them correctly. These conventions are called protocols. On the
Internet there exists a hierarchy of protocols for different tasks in
the process of data transmission. In the following we will introduce
IP and TCP, the two most basic protocols.
The Internet Protocol (IP) controls the transfer of small packages of
data on the net. Therefor the packages are labelled with an address
and then send to the corresponding computer on the Internet.
An Internet address is composed of four numbers each less than 256.
See for example the address of a computer named ente at the
University of Bielefeld:
129.70.133.27
The beginning of an address denotes the network a computer is part of
while the end of an address denotes the computer itself.
If data should be send to a computer that is not part of the local
net, it has to be "routed". Each router on the Internet is connected
to several other routers, hence, packages of data are send from router
to router (each of them closer to their destination) until they reach
their destination network.
The Transmission Control Protocol (TCP) is used to transmit larger
packages of data.
TCP splits up the information into smaller units which are numbered
and supplied with a checksum, then, these units are send off using IP.
The receiving computer puts the units back in the proper order and
verifies by means of the checksum whether they were transmitted
correctly. If a package of data is missing or was not transmitted
correctly, it is asked for once again.
Figure 3
sketches the hierarchy of protocols.
In order to identify packages of data the TCP header also contains
port information. E-mail for example has port information
25 while
telnet has port information 23 .
If a computer receives data at port 25 , a mail daemon is started
which is a program that handles incoming e-mail.
Thus, thanks to the port information it is possible to keep different
network applications working simultaneously without mixing up the
received data.
As we have already seen, to be identified each computer on the
Internet needs a unique address which is composed of four numbers.
However, most people prefer to remember names than to retain many
different numbers. Consequently, apart from an address each computer
on the net also has a name and the Domain Name System (DNS)
ensures that no other computer on the Internet has the same name:
The Internet is subdivided into different domains which again are
subdivided and so on. For every domain there is a group of people
responsible for the unambiguity of the names denoting the subdomains.
The computer with address 129.70.133.27 for example has the
name:
ente.techfak.uni-bielefeld.de
We can see that ente is part of the domain techfak
(Faculty of
Technology) which again is part of the domain named uni- bielefeld
(University of Bielefeld). The highest level domain is de
(Germany).
Originally there were six highest level domains: com, edu, gov, mil
org,
and net . They denoted commercial, educational, governmental
and military organisations in the USA, other organisations and Internet
specific institutions.
It must be mentioned that in general domain names do not correspond
to geographical locations: A computer situated in Germany may be part
of the domain de as well as of the domain com or any
other.
If a computer wants to contact another computer given its name, first
of all the corresponding address has to be determined. There are three
possibilities to get this address:
- The address can be found somewhere in the memory, because the other
computer was just contacted.
- The address is known, because the other computer is part of the local
net.
- If the address is unknown, a Root-Server has to be contacted. This is
a computer that knows the addresses of all highest level domain
Name-Servers which are computers where the names of corresponding
subdomains are stored. The Root-Server contacts a subdomain
Name-Server which again contacts a sub-subdomain Name-Server.
This process continues until a Name-Server is reached which is
part of the same local net as the computer that originally was
intended to be contacted. This Name-Server possesses the desired
address and sends it back to the requesting computer.
As an example Figure 4
sketches how the computer
ente.techfak.uni-bielefeld.de
contacts a computer named
pollux.inrialpes.fr (194.199.21.144)
which obviously is not part of ente's local net.
back to
the previous section
or continue with
the next section