APNIC chief scientist: IPv6 introduction probably obsolete

If you talk to networkers about IPv6, there is often an awkward silence: The successor to the IPv4 address standard dates back to 1998 – and has not yet replaced IPv4 across the board, despite all attempts by the network community. While it is generally good practice in the network environment to build sensible dual-stack setups with support for IPv4 and IPv6, many providers are relaxed about the issue: Large sites such as GitHub still do not support IPv6. In this context, there is often talk of the chicken-and-egg problem, as ISPs do not offer IPv6 because there is allegedly not enough demand from customers – after all, many sites on the Internet are not accessible via IPv6. Sites with an IPv4 setup, on the other hand, claim that there is no point in introducing IPv6 as long as potential visitors cannot use the standard anyway due to a lack of their own IPv6 configuration. In any case, IPv6 is still a long way from the declared goal of putting an end to the foreseeable shortage of IPv4 addresses by increasing the address space. A lively trade in IPv4 addresses has long since been established.

Against this backdrop, an interview with the chief scientist of APNIC, the IP allocation body responsible for the Asia-Pacific region, caused quite a stir: in it, Geoff Huston stated that the goal of completely replacing IPv4 with IPv6 was obsolete in his eyes.

Names instead of addresses

According to Huston, there are many reasons for this. Firstly, a major problem was that the need for IPv6 could not be meaningfully communicated to many Internet users because it did not offer any additional functions apart from the larger address space. To a certain extent, there was no incentive to switch to IPv6. At the same time, the planned introduction of IPv6 has, unfavorably, also led to the use of mobile devices such as iPhones, which have presented providers with completely new challenges in terms of traffic. According to the theory, companies that might otherwise have been concerned with the introduction of IPv6 were busy adapting their environments to the growing demand from mobile devices. In doing so, they relied on various types of NAT, for example, which still works with sufficient quality for most users today.

Although IPv6 has now reached a penetration rate of over 40 percent of Internet users, Huston attributes this primarily to the fact that only relatively small IPv4 address blocks are reserved for the APAC region as a whole. IPv6 is particularly important there. However, Huston believes that another factor indicates that the train for IPv6 may have left the station: the industry has established the principle of content delivery networks (CDNs) in order to deliver the ever-increasing amount of content to users. For these, however, names are primarily important and not so much the IP addresses used. Even today, a large part of access control in the network is mainly carried out via DNS and load balancers. The importance of IPs is therefore fading into the background, and the central magic has to happen in the respective applications instead.

Accordingly, Huston advocates a more pragmatic approach to the migration to IPv6: the goal should no longer necessarily be to completely replace IPv4 with IPv6. Instead, one could claim that the IPv6 transition has been successfully completed if it is possible for ISPs to serve their own customers without IPv4. It all depends on the details: Even for Huston's stated goal, it would ultimately be necessary for all ISPs as well as all service providers to support IPv6. Whether IPv4, IPv6 or even completely different technologies are used in the background of applications would ultimately be irrelevant according to this logic.

In his conclusion, Huston even goes one step further: he questions the definition of the Internet as a "network of networks" with a shared address space and assumes that in future, the Internet will be seen more as a network of services that use similar access methods. In general, the relevance of the network will continue to diminish and the importance of services will increase. Interested readers can find the full article on the APNIC blog.

Finger in the wound

Whether Huston's predictions will come true is, of course, written in the stars. In any case, his interview puts his finger in the IPv6 wound described at the beginning. Given the very slow pace of IPv6 adoption throughout the network community, it is at least understandable that some people no longer believe that IPv6 will finally and completely replace IPv4. However, Huston's comments should not be overestimated – Anyone who wants to use the interview as an excuse to put their own project for the introduction of IPv6 on ice would be wrong: After all, even Huston does not question the fundamental necessity of introducing IPv6 in a functioning manner throughout the entire network. And the Internet is currently still a long way from the goal of having competitive ISPs without IPv4.

(nen)