LPI Linux Certification in a Nutshell (58 page)

For several years IPv4 has been the standard method for
assigning a unique address that identifies the host on the network and
the Internet. The 32-bit IP address, also referred to as a
dotted quad
, is composed of four 8-bit fields
divided by a period. These fields identify first the network and then
the host for a device on the network. IPv4 provides 4.29 billion
addresses.

The IP address 192.168.1.150 and the binary equivalent would
be:

IPv4 addresses are categorized into classes to provide structure.
There are five classifications of networks defined by IP addresses.
Table 19-1
identifies the address
ranges for the classes that are primarily used.

Every device on a network or connected to the Internet needs a
unique IP address, including printers and fax machines; consequently,
the supply of available IP addresses will run out eventually. There have
been many technologies developed to reduce the exhaustion of available
IP addresses, including network address translation (NAT), Classless
Inter-Domain Routing (CIDR), and IPv6.

The early specification of the IP recognized that it would
be necessary to divide one’s given allotment of IP addresses into
manageable subnetworks. Such division allows for distributed management,
added security (fewer hosts can potentially snoop network traffic), and
the use of multiple networking technologies (Ethernet, Token Ring, ATM,
etc.). IP also enables convenient partitioning of the physical portions
of a network across physical and geographical boundaries. To provide the
capability to locally define networks, IP addresses are considered as
having two distinct parts: the part that specifies a
subnet
and the one that specifies a
network interface. (Remember that IP addresses are
assigned to network interfaces, not host computers, which can have
multiple interfaces. For this discussion, however, we assume a
one-to-one relationship between hosts and interfaces.) The boundary
between the network and host portions of an IP address is delineated by
a
subnet mask
, required by the TCP/IP configuration
of any network interface. Like the IP address, the subnet mask is simply
a 32-bit number specified in four 8-bit segments using
dotted
quad
decimal notation. The familiar class A, B, and C
networks have these subnet masks:

When logically AND’d with an IP address, the bits set to
0
in the subnet mask obscure the host portion
of the address. The remaining bits represent the network address. For
example, a host on a class C network might have an IP address of
192.168.1.127. Applying the class C subnet mask 255.255.255.0, the
network address of the subnet would be 192.168.1.0, and the host address
would be 127, as depicted in
Figure 19-1
.

Figure 19-1. Host interface address calculation

Although it is typical to use the predefined classes (A, B, and
C), the boundary can be moved left or right in the IP address, allowing
for fewer or more subnets, respectively. For example, if a single
additional bit were added to the class C subnet mask, its IP address
would be:

With such a subnet defined on an existing class C network such as
192.168.1.0, the 256-bit range is split into two subnets, each with 7
host bits. The first of the two subnets begins at 192.168.1.0 (the
subnet address) and continues through 192.168.1.127 (the subnet
broadcast address). The second subnet runs from 192.168.1.128 through
192.168.1.255. Each of the two subnets can accommodate 126 hosts. To
extend this example, consider two additional bits:

When applied to a class C network, four subnets are created, each
with 6 host bits. Just as before, the first subnet begins at 192.168.1.0
but continues only through 192.168.1.63. The next subnet runs from
192.168.1.64 through 192.168.1.127, and so on. Each of the four subnets
can accommodate 62 hosts.
Table 19-3
shows more detail on class C subnets, considering only the host portion
of the address.

As you can see, as the number of subnets increases, the total
number of hosts that can be deployed within the original class C address
range reduces. This is due to the loss of both broadcast addresses and
network addresses to the additional subnets.