Each NIC must be a part of an IP network. An IP network has an address, and a NIC has an address as well. In IP version 4 (IPv4), a NIC has an IP address that is a 32-bit number.
An IP network address specifies the most significant n bits of the IP addresses of hosts in the network. A host in an IP network supplies the additional 32 − n least significant bits for the complete 32-bit IP address.
For example, let us consider an network address of 110000001010100000000001000000002. The subscript 2 means this is a binary (base-2) number. In this notation, it is not clear how many of the least significant bits can be specified by a host (NIC).
A netmask is a bit pattern that is used to designate the network address portion versus the host portion. In our example, 111111111111111111111111000000002 is a netmask that says the most significant 24 bits are the network address, while the least significant 8 bits form the host address. In other words, in a netmask, a ‘1’ means it is a part of the network address, a ‘0’ means it is a part of the host address.
In this example, it appears that there should be 28 = 256 possible host addresses. This is not true, however. Several special host address bit patterns are reserved. If all the host address bits are zeros (0), then the address is the network address. No host (NIC) may assume the network address. If all the host address bits are ones (1), then the address is called a broadcast address, which will be explained later.
When the network address and the host address are combined, it is called an IP address.