The Internet Assigned Numbers Authority (IANA) announced in February 2011 that its pool of IPv4 addresses1 was finally exhausted. The doom merchants had pretty much forecast the end of civilisation when this event occurred if we had not already started the transition to IPv6 with its greatly extended address capacity, but in reality life has carried on much as normal. So why has this apparently catastrophic event suddenly become less of an issue? The reality is that the next tier down in the address allocation hierarchy, termed the Regional Internet Registries (RIRs) – of which there are five in total2 – hold their own address pools, as do the ISPs they serve, which both act as a further buffer of spare addresses.
Figure 1: Simplified representation of address allocation hierarchy [Source: Analysys Mason, 2012]
The forecast exhaustion point for the RIR pools varies, with the pool of IPv4 addresses for the Asia-Pacific region already exhausted, whilst that of the Africa region is forecast to be exhausted by October 2014. So whilst the IANA IPv4 address exhaustion date has passed without any calamitous results, the date by which ISPs can no longer allocate new IPv4 addresses is rapidly approaching and that is the point where the real problems start.
The issue of IPv6 readiness has therefore never been so important. The majority of ISPs now have IPv6 allocations, or they can readily obtain them from their RIR on demand. The vast majority of equipment vendors now supply dual-IPv4/6 working devices and similarly operating systems are capable of dual-stack working: e.g. Windows XP/Server 2003 onwards can support IPv6. The IPv6 Forum runs a programme for the testing and approval of IPv6 devices under the ‘IPV6 Ready Logo Program’ – the supply side is therefore broadly ready to support IPv6.
Despite the impending exhaustion of IPv4 addresses, the levels of IPv6 readiness remain worryingly low in some end-user sectors. We know for instance from a study that Analysys Mason undertook recently in Singapore3 that knowledge levels about IPv6 are particularly low in the SME sector, which, bearing in mind this sector accounts for the largest proportion of the working population, suggests more publicity is required to raise awareness levels. This represents an early challenge that the many IPv6 taskforces that are being set up around the world need to address.
The proportion of IPv6 traffic across the World Wide Web remains tiny compared with that of IPv4 (less than 0.03% in 2011)4. However, there are an increasing number of examples of organisations adopting IPv6 for certain applications, for instance tunnelling across the web between different international locations. Google and Yahoo now provide IPv6 services and we expect an increasing number of organisations to launch IPv6 services in the near future.
The key question for most organisations will be when IPv6 should be implemented. Our view is that no simple answer exists, but it is important that every organisation has at least a plan. We recommend starting with an audit of existing infrastructure, operating systems and applications to determine the level of IP readiness and identify where problems for IPv6 migration may exist, such as legacy hardware or systems that cannot be upgraded.
The next step is to consider the needs of the organisation: e.g. is it likely to require large amounts of new address space in the foreseeable future, or maybe users will need to start accessing IPv6 websites. It is worth remembering that countries like China will lead the curve in launching IPv6 sites and similarly for users being allocated IPv6 addresses, so if these are potential trading partners for instance, some early action will be required. This activity can help with assembling a business case for commencing the transition work and identifying if the business is exposed to any risk.
It is generally accepted that most organisations will need to begin with a period of dual-stack working, i.e. running both IPv4 and IPv6, which will then run for several years. The final transition to a wholly IPv6 environment is not expected to occur until 2020 at the earliest, and possibly much later according to other forecasts.
Certain applications could drive the early adoption of IPv6 by some organisations, either as a future-proofing measure during the implementation of new systems, or because the ability to assign unique IP addresses (i.e. not re-used repeatedly as now needs to happen with IPv4) offers specific advantages. For instance, IPv6 could be immediately used for new utility telemetry applications where there is no interaction with other external networks. Other examples include fourth-generation LTE mobile networks that will use voice over IP and therefore place a heavy demand on IP address space, similarly for M2M applications such as smart meters.
1 IPv4 was launched in the 1970s as the 1st generation IP addressing scheme and comprises 32-bit numbers that are often expressed as 4 octets in "dotted decimal" notation (e.g. 192.0.2.53). The IPv6 protocol was actually first developed in the mid-1990s when it was identified then that the available IPv4 address pool could not cope with explosive growth in address demand driven by Internet usage. IPv6 addresses are 128-bit numbers and are conventionally expressed using hexadecimal strings (e.g. 2001:0ba0:01e0:d001:0000:0000:d0f0:0010).
2 RIRs comprise: AfriNIC (Africa and parts of the Indian Ocean); APNIC (Asia/Pacific Region); ARIN (North America and parts of the Caribbean); LACNIC (Latin America and parts of the Caribbean); RIPE NCC (Europe, the Middle East and Central Asia).
3 Study available from http://www.ida.gov.sg/Technology/20110414110718.aspx
4 http://ddos.arbornetworks.com/2011/06/world-ipv6-day-final-look-and-wagons-ho/