This is an archive page of an article written by the RIPE NCC in 2003.

IPv4 - Running Out of Time?

News reports claim we will soon run out of IP addresses. Some reports say that fast growing economies (such as those in Asia) look set to be the ones first hit. America and Europe will be next. Some claim African nations have been starved of IP addresses.

The RIPE NCC, along with other Regional Internet Registries (RIRs) report that this is not based on solid, publicly available statistics. Analysis of trends based on early data is not always the best way forward. Past trends may not lead to future change.

The challenge facing the Internet is based on the fact that the way of handing out IP addresses was set up in the early 1980s. The current system – known as IPv4 - provides for around 4.4 billion unique IP addresses. When this system was devised, this was thought to be more than enough. The Internet grew larger than anyone expected. The current estimate is that there are more than 100 million hosts and 350 million users actively on the Internet.

IPv4 uses addresses with two parts - ‘a network identifier’ and ‘a host identifier’ within that network. Initially, just the first octet of the number identified the network; the other three were used to identify the host device.

This idea was soon refined, and the idea of a ‘class-based address architecture’ was born. One half of all of the available address space was left as the initial 8/24 bit structure and called the Class A space. This allowed for up to 127 networks each with 16,777,216 host identities – the BIG networks that were planned when the system was devised.

There also had to be some provision for medium sized and small networks. One quarter of the remaining space was split into 16/16 bits (allowing for up to 16,128 networks, each with up to 65,536 hosts) called Class B space. Class C space was given a further one eighth of the remaining space. This was divided using a 24/8 bit structure, (allowing for 2,031,616 networks, each with up to 256 hosts). The remaining one eighth of the space was held in reserve.

It became clear, early in the 1990s that space might one day run out. At the heart of the shortage was the way in which it had been split. The Class A networks were too large, the Class C networks were too small. Network administrators began to ask for Class B addresses, giving themselves built-in room for growth. As a result, there was much wasted space. It seemed that once it was gone, there was no way of getting it back. Figures suggested that 17% of the total address space was allocated but not announced to the Internet.

As a short-term measure, the Internet Engineering Task Force (IETF) took on the task of finding some answers. They came up with a number of answers.

The IETF relaxed the ‘class system’. This cut down wasted space. The network identifier could be any length at all. It was now the couplet of an IP address prefix and the bit length of the network part. The boundary between the network and host part could change across the network. This was called the "Classless" address architecture, also known as Classless Interdomain Routing (CIDR). It allowed for better use of the remaining space.

The IETF also devised Network Address Translation (NAT). A single device, such as a router, acted as an agent between the Internet (or "public network") and a local (or "private") network. An entire group of computers needed only a single, unique IP address. Many engineers dislike the idea of NAT. They feel IPv6 will return the Internet to its original aims, based on peer-to-peer connections. NAT is often blamed for falling service levels across the Internet. Poorer performance lowers expectation and could limit the credibility of the Internet.

The Internet Assigned Numbers Authority (IANA) looks after the remaining global pool of around sixteen million IPv4 addresses. IANA allocates address blocks to RIRs as they are needed. By the middle of 2003, research showed that of the 221 blocks, 90 remained unallocated by the IANA.

The Internet is still growing quickly. RIRs allocated around 4.25 blocks of address space in 2002 and 5.5 blocks in 2001. This suggests that that there is not a looming shortage in the IPv4 address space. Even allowing for a dramatic increase in address consumption rates, it is likely that IPv4 address space will last well beyond the two years some predict.

The RIRs handle address space allocation, guided by IANA policies. When IPv4 address space finally "runs out" this will occur at a global level. Each region will have a small pool of addresses left to allocate. Addresses are distributed in a co-ordinated fashion from one single global pool. The pool is not pre-allocated to different countries or regions. The claim that Asia will be first to lose out is simply untrue.

Estimates of an exhaustion date assume no major technology or policy changes. Any change in the distribution system might cause disruption.

The current RIR and LIR distribution model has limited the amount of address space in idle holding pools. There has been a suggestion that we should follow the models used in other areas, such as that used by telephone companies. This distribution system relies on a geo-political structure. Number blocks are passed to national bodies; further allocation is a matter of local policy. This system would increase the number of 'holding' points in the distribution system. It locks away larger pools of address space from being deployed and advertised. It is likely to lead to an upward change in the overall address usage rates due to an increased inefficiency.

The general feeling is that the figures we can usefully use span from the mid to late nineties. Looking at how take up of address space has progressed until now, experts are divided on when addresses will run out. Estimates of this time fall around 2012.

This means that we do not have ‘forever’. Plans are in place to find a new way of allocating addresses well ahead of the pool running out.