IP addresses in these ranges are not routable in the
Internet. When private networks using these IPv4
addresses need to connect to the Internet, their
addresses need to be translated to public, Internet-
routable IPv4 addresses. This translation process is
called Network Address Translation (NAT), and it is
discussed in more detail in Chapter 18, “IP Services.”
The need to further divide IPv4 address classes,
especially the Class A and Class B networks, and to create
additional subnetworks became clear as networks started
to grow in size and complexity and IPv4 addresses
started to run out. As previously mentioned, a Class A
network with the default address mask can address more
than 16 million devices on the same network. Having
more than 16 million devices on one Layer 3 network is
theoretically possible but practically impossible. Class B
networks have 16 bits dedicated to host IDs, which
means each Class B network can address 2 = 65,536 − 2
= 65,534 devices (with two addresses reserved for the
network address and the broadcast address). These
networks also are in need of further subnetting.
Subnetting is the process of borrowing bits from the host
ID portion of a network and transforming them into
network ID bits.
Classless interdomain routing (CIDR) goes beyond the
class limits and the default address masks and represents
a way of specifying the subnet mask for an IP address.
Initially when the classes of IPv4 addresses were defined,
the default masks for each class were already implied
and were not even mentioned when assigning networks
for customer consumption. As the available IPv4
addresses became more and more scarce, especially for
Class A and Class B networks, CIDR was introduced as a
way to capture the transition from a class-based Internet
to a classless one. In a classless Internet, it is no longer
feasible to assume that the address mask for a Class A
network is 255.0.0.0 or /8 or for a Class B network that
16