IP Address Space Depletion

The growing demand for IP addresses has put a strain on the classfull model, especially class B address space, which was getting depleted at a fast pace. Most companies requesting IP addresses have estimated that a class B would meet their requirement because it is a fair balance between the number of networks and the number of hosts. A class A was overkill with more than 16 million hosts, and a class C had too few hosts per network. By 1991, it was becoming obvious that the class B consumption was not slowing down and actions needed to be taken to prevent its depletion. Some of these measures consisted of creative assignment of IP addresses and promoting the use of private IP addresses for organizations that do not have global connectivity needs. Other measures resulted in the initiation of working groups and directorates such as the Routing and Addressing (ROAD) working group and the IP next generation (IPng) directorate. In 1992, the ROAD working group proposed the use of Classless Interdomain Routing (CIDR) as a measure to move away from classfull IP addressing. At the same time, the IPng directorate was working on developing a new and improved IP addressing scheme, IP version 6 (IPv6), that would eventually solve the problems that IPv4 is encountering.

The measures to handle the IP depletion can be grouped in the following categories:

•  Creative IP address space allocation

•  Classless Interdomain Routing (CIDR) [4]

•  Private addressing and Network Address Translation (NAT) [6,7]

•  IP version 6 (IPv6) [8]

Along with depletion concerns, growing IP address demand generated a need to convert the IP addressing allocation process from a central registry. Originally, the Internet Assigned Numbers Authority (IANA) and the Internet Registry (IR) had total control for address assignment. IP addresses were assigned to organizations sequentially without any consideration of geographical factors and to how or where an organization would plug in into the Internet. This method had the effect of punching holes in the IP address space; that is, segregating individual or small numbers of IP addresses and eliminating large, contiguous ranges of numbers.

A different approach needed to be taken whereby a large, contiguous range of addresses is given to different administrations (such as service providers), and those providers in turn allocate addresses from their own space. In general, this funnel-down method of address allocation predicts a more controlled and hierarchical method of IP address distribution.

IP Address Allocation

Class A network numbers are limited resources, and the allocation from this space is restricted. Although the upper range of class A 64 to 127 will be distributed, there is still no definite plan on how this is going to be done.

Class B addresses are also restricted. They will be allocated only if the need for such addresses is fully justified. Due to the scarcity of class B network numbers and the under utilization of the address space by most organizations, the recommendation is to use multiple class Cs instead.

The class C network number space is now being divided and allocated in a way that is compatible with address aggregation techniques. Address aggregation is the practice of summarizing a contiguous block of addresses in a single notation, or advertisement. (Aggregation is also relevant to the CIDR model, to be discussed in the next section.)

Class C addresses are being distributed to ISPs with the requirement that the original allocation for the provider should last at least two years. In turn, each provider will allocate a block of addresses from its own range to each of its customers. Customers will not be granted more addresses from their ISPs until 80 percent of the original address space granted to the customer has been used. The allocation of class C addresses is illustated in table 3-1.

If a subscriber has a requirement for more than 4,096 IP addresses, a class B network number might be allocated.

Organizations are encouraged to use Variable Length Subnet Mask (VLSM) as much as possible to use the address space more efficiently.

As far as the geographic allocation of blocks of C addresses, there are four major areas: Europe, North America, the Pacific Rim, and South and Central America. The allocation is summarized in table 3-2. The multiregional area represents network numbers that have been assigned prior to the implementation of this plan. Ranges designated as "Others" are for geographical areas other than the areas named specifically.