Intro' to the TCP/IP Protocol Suite.

Preamble

As noted before, tcp/ip uses its own addressing scheme. This means that the application layer only has to deal with one single address format knowing that the network layer will convert it to the appropriate physical address. This signifigantly reduces the complexity of the Application layer, while shifting the burden to a layer closer to the physical network.

IP Addresses

Each host on a network is assigned a unique address from the IP address space.

IP addresses are 32 bits long, and normally represented as 4 decimal numbers seperated by periods (7F:00:00:01 or localhost, is represented as 127.0.0.1).

IP addresses are divided into 3 sections. The class, the network id, and the host id.

Class A : 0 , 7-bit netid, 24-bit hostid.
Class B : 10 , 14-bit netid, 16-bit hostid.
Class C : 110 , 21-bit netid, 8-bit hostid.
Class D : 1110 , 28-bit multicast address.
Class E : 11110, Reserved.

Note that there are only three classes available for general use. Also note that this approach to dividing the address space is no longer strictly used. It does however provide a good reference, and the terms Class A, B, C are still widely used to describe address spaces.

Because IP Addresses contain both a network and a host id, they do not specify a computer. Rather they specify a network connection.
This may seem to be of small consequence. However when you start to deal with routing issues and multi-homed hosts, it ceases to be an academic point.

Some Addresses have been reserved for special purposes.
A host id of all ones, specifies a broadcast to all hosts on the current network (note that the network is defined by the netid, NOT by the physical layout).
A host id of all zeros, refers to the network itself. So a class A network with netid 0A, is refered to as 10.0.0.0.
A netid of zero, refers to the current network. This is not often used, and is normally restricted to system startup when the host may not know it's netid.

<< Contents >>

Andrae Muys