CIDR swapped fixed address classes for prefix lengths so the net could scale

The classful plan offered three sizes: tiny Class C, huge Class B and enormous Class A. Middle-sized organizations had no good fit, so Class B, the best available, was being exhausted while routing tables exploded.

CIDR deprecates the classes in favor of a prefix length. A block can be any size, and a provider can hand customers addresses from its own prefix, so all their routes collapse into a single advertisement.

The design that made it inevitable was topology-based assignment plus aggregation: the more everyone adopted it, the smaller the routing table, so the pressure to adopt grew instead of shrinking.

Replace fixed classes with variable prefixes

Variable prefixes fit the block to the need, conserving scarce address space.

Hierarchical assignment lets providers aggregate routes, shrinking routing tables.

It needed no new hardware or protocol; it changed how addresses were assigned and advertised.

When a fixed taxonomy is exhausting a scarce resource, introduce variable granularity so you can match supply to need and summarize what you hand out.

CIDR, standardized in RFC 1519 and updated in RFC 4632, kept IPv4 viable through the 1990s and 2000s and remains the basis for address allocation and route aggregation. It did not solve IPv4 exhaustion permanently; that task fell to NAT and eventually IPv6.