Why did CIDR replace IPv4 address classes?

The historical model

Classes made the first network boundary predictable.

Classful IPv4 addressing used leading bits to select a default network size. Class A used a default /8 network, Class B a /16, and Class C a /24.

That simple model became wasteful as networks grew in many different sizes. An organisation needing more than a /24 but far less than a /16 could not choose a right-sized classful block.

Classes are historical terminology; CIDR is the operational model.You may still hear “Class C” in older documentation, but modern routing and allocation use explicit prefix lengths such as /20 or /27.

Build the model

Five historical ranges

Class A

First octet 1–126 in the traditional usable range; default /8 network.

Class B

First octet 128–191; default /16 network.

Class C

First octet 192–223; default /24 network.

Class D

224–239, reserved for IPv4 multicast groups rather than host subnets.

Class E

240–255, reserved for experimental or future use in the historical model.

Ranges are not permissions to assign an address

An Internet Protocol version 4 (IPv4) address has 32 bits, grouped into four 8-bit octets. Historically, leading bits 0, 10, 110, 1110 and 1111 identified classes A, B, C, D and E. Class A's leading-bit range includes first octets 0 and 127, but these have special uses; 127.0.0.0/8 is loopback.

Class D is the multicast range 224.0.0.0/4, not a unicast subnet class. The Internet Group Management Protocol (IGMP) manages IPv4 multicast membership. Class E is the historical reserved range 240.0.0.0/4; it includes 255.255.255.255, the limited-broadcast address. Neither range is an ordinary unicast allocation pool. Private use is defined by reserved prefixes, not by A/B/C labels.

Compare with CIDR

Prefix length gives the network the right size

CIDR separates the network prefix length from the old class labels. A /20 contains 4,096 IPv4 addresses, while a /27 contains 32, so allocation can better match the need and routes can be aggregated.

Historical IPv4 classes A through E compared with CIDR blocks using /20 and /27 prefixes. — Classful ranges were fixed; CIDR chooses an explicit prefix length.

Count the block, then check its use

For an IPv4 prefix of length p, the block contains 2^(32 - p) total addresses. In a conventional subnet, two addresses identify the network and directed broadcast, leaving two fewer host addresses. Do not apply that subtraction blindly: /31 point-to-point links and /32 single-address routes are special cases.

CIDR also supports route aggregation. The aligned blocks 10.20.0.0/24 and 10.20.1.0/24 fit into 10.20.0.0/23. Contiguous blocks must fit the aggregate boundary; reachability and routing policy must also support the advertised route. An aggregate does not prove that every address is reachable.

Compare one subnet for 300 interface addresses

Compare capacity, not physical size. Assume one conventional IPv4 subnet with network and broadcast addresses reserved; the 300 required interface addresses include any gateway interface.

Knowledge check