Classless Internet Domain Routing
Class A addresses are long gone, and the world is quickly running out of Class B addresses. Many Class C addresses are still available, but the small address space of a Class C network (254 hosts maximum) is a severe limitation in the high-volume game of Internet service providers (ISPs). It is possible to assign a range of Class C networks to a network owner who needs more than 254 addresses. However, treating multiple Class C networks as separate entities when they are all going to the same place only clutters up routing tables unnecessarily.
Classless Internet Domain Routing (CIDR) is a technique that allows a block of network IDs to be treated as a single entity in routing tables. CIDR groups a range of network IDs into a single address entry using what is called a supernet mask. You can think of a supernet mask as something like the opposite of a subnet mask. Instead of designating additional bits for identifying the network, the supernet mask in effect takes bits away from the network ID. The addresses in the range are therefore identified by the network address bits that the networks in the range hold in common. For example, an ISP might be assigned all Class C addresses in the range
to 220.127.116.11 (11001100000101011111111111111111).
In this case, the network addresses are identical up to the seventeenth bit counting from the left. The supernet mask would therefore be 11111111111111111000000000000000, which is equivalent to the dotted decimal mask 255.255.128.0.
The address block is specified using the lowest address in the range followed by the supernet mask. A common notation for a CIDR address/mask pair is to show the number of mask bits after the address with a slash (/) separator. Hence, the CIDR range in the preceding example would be specified as 18.104.22.168/17.