CIDR(7)CIDR(7)NAME
CIDR, cidr - Classless Inter-Domain Routing
DESCRIPTION
Classless Inter-Domain Routing, also known as CIDR, is an Internet
addressing architecture designed to solve two problems: the growth in
the size of the routing tables in the top-level routers and the exhaus‐
tion of Class B address space. To solve these problems, CIDR relies on
a new means of distributing the allocation of Internet address space
and on a concept known as route aggregation.
For an up-to-date list of Frequently Asked Questions (FAQ) about CIDR,
retrieve the CIDR FAQ from either of the following locations:
http://www.rain.net/faqs/cidr.faq.html
http://www.ibm.net.il/~hank/cidr.html
Address Space Allocation
The traditional 32-bit Internet address is divided into a network part
and a host part. The size of each part depends on the network class to
which the address belongs. The following table shows the sizes of each
part for Class A, Class B, and Class C networks.
─────────────────────────────────────────────────────────────
Network Size of Network Part Size of Host Part (in bits)
(in bits)
─────────────────────────────────────────────────────────────
Class A 8 24
Class B 16 16
Class C 24 8
─────────────────────────────────────────────────────────────
In this scenario, each physical network or LAN uses a single network
number. While the idea seems sound, most organizations seldom create a
single network containing thousands of hosts, choosing instead to
divide their networks using routers. For organizations with Class B
addresses, this wastes valuable addresses.
To cope with the various network topologies, the concept of subnetworks
or subnetting emerged. In a subnet, the network part of the address
consists of the network part and a portion of the host part. The bit‐
mask convering these two parts is called the subnet mask. The area of
the host part that is covered by the subnet mask identifies the subnet.
This process allows you to identify individual LANS by their subnet
number within the larger network number. The only way to communicate
between two or more subnets is through a router.
Currently, routers make routing decisions by extracting the network
portion of an IP address and looking it up in their routing table. This
forces some IP routers to store each network number connected to the
Internet in their routing table.
For many organizations, a Class C network (254 hosts) is too small,
whereas a Class B network (65534 hosts) is too large, resulting in poor
address space utilization.
Route Aggregation
The Internet Advisory Board (IAB) and Internet Engineering Task Force
(IETF) have decided to eliminate the notion of IP address classes and
to direct routers to make routing decisions based on a variable-length,
contiguous IP address prefix. This is what is meant by classless rout‐
ing. Under this scenario, an Internet Service Provider (ISP) that had
previously announced 256 contiguous Class C networks to the Internet,
now only has to announce a single prefix, with 16 significant bits, for
all these networks. This prefix is referred to as an aggregate, and the
network is referred to as a supernet. If the ISP needed to add addi‐
tional customers to its network, it could do so without modifying the
routing announcements to the rest of the Internet.
Aggregating networks reduces the number of routers in a network and
enables you to make optimum use of bridges and high-speed switches.
EXAMPLES
This section describes one example of a Class C supernet. If organiza‐
tion A requires 1000 addresses, it might have the following Class C
networks: 212.221.32.0, 212.221.33.0, 212.221.34.0, and 212.221.35.0.
Using current Class C addressing specifications, organization A's net‐
work mask and network numbers are as follows:
/------------24 bits-----------\ 1111 1111 1111 1111 1111 1111 0000
0000 = mask 255.255.255.0 1101 0100 1101 1101 0010 0000 0000 0000 =
network 212.221.32.0 1101 0100 1101 1101 0010 0001 0000 0000 = net‐
work 212.221.33.0 1101 0100 1101 1101 0010 0010 0000 0000 = network
212.221.34.0 1101 0100 1101 1101 0010 0011 0000 0000 = network
212.221.35.0 \--------network address-------/ \--host--/
address
Typically, software compares all network address bits that are covered
by the network mask (1 bits) to determine the effective network
address. Because the network addresses covered by the 24-bit network
mask are different, traffic from one network to another requires a
router. In addition, routes to each of the four networks are adver‐
tised to the rest of the Internet, and occupy space in the routers'
routing tables.
Under CIDR rules, organization A could shorten their network mask from
24 bits under current rules to 22 bits. The result is a network mask of
255.255.252.0, as follows:
/-----------22 bits----------\ 1111 1111 1111 1111 1111 1100 0000
0000 = mask 255.255.252.0 1101 0100 1101 1101 0010 0000 0000 0000 =
network 212.221.32.0 1101 0100 1101 1101 0010 0001 0000 0000 = net‐
work 212.221.33.0 1101 0100 1101 1101 0010 0010 0000 0000 = network
212.221.34.0 1101 0100 1101 1101 0010 0011 0000 0000 = network
212.221.35.0 1101 0100 1101 1101 0010 0011 0000 0000 = network
212.221.35.0 \-------network address------/\----host----/
address
Because the network addresses covered by the 22-bit network mask are
the same, traffic from one network to another does not require a
router. Instead, the software uses Address Resolution Protocol (ARP) to
acquire direct connection to the network.
The address 212.221.32.0 with the mask 255.255.252.0 identifies all
networks belonging to organization A. Expressed in CIDR format, orga‐
nization A's network address is 212.221.32.0/22. This effectively
aggregates all routes under one network address. This also means that
only one route is advertised to the rest of the Internet. If a router
sees traffic addressed to 212.221.33.5 with the netmask of
255.255.252.0, the traffic is addressed to network 212.221.32.0.
Using a network mask of 255.255.252.0, organization A can have a single
bridged network of 1022 hosts (hosts 0 and 1024 are reserved for the
broadcast address). Using a network mask of 255.255.254.0, organization
A can have two bridged networks of 510 hosts (host 0 and 512 are
reserved for the broadcast address).
These techniques are not currently implemented in all host software,
and should be implemented in networks with great care. However, the
IETF suggests that host software be modified to allow for classless
routing.
SEE ALSO
Commands: netstat(1), ifconfig(8), route(8)
RFC1517, Applicability Statement for the Implementation of Classless
Inter-Domain Routing (CIDR)
RFC1518, An Architecture for IP Address Allocation with CIDR
RFC1519, CIDR Address Strategy
RFC1520, Exchanging Routing Information Across Provider Boundaries in
the CIDR Environment
CIDR(7)