Today the Internet Governance Project published a new report indicating that the transition to a new Internet standard may get stuck somewhere between the old and the new.
The data communications protocol supporting the Internet (IPv4) is almost 40 years old, and its 32-bit address space is too small for the global Internet. A new, “next generation” Internet Protocol (IPv6), has a much larger, 128 bit address space. But the new protocol is not backwards compatible with the existing Internet. For the past 20 years, the Internet technical community has been trying to migrate the entire Internet to the new standard.
The study, titled “The Hidden Standards War: Economic Factors Affecting IPv6 Deployment,” addresses important but often overlooked questions about the technical evolution of the Internet: Will the world converge on IPv6? Will IPv6 die out? Or will we live in a mixed world for the foreseeable future? The study’s authors are Dr. Brenden Kuerbis, a postdoctoral researcher, and Dr. Milton Mueller, a Professor in Georgia Tech’s School of Public Policy. Their independent research was supported in part by the Internet Corporation for Assigned Names and Numbers (ICANN)’s Office of the Chief Technology Officer.
The research offers a clear-eyed, economically-grounded study of IPv6’s progress and prospects. Many promoters of IPv6 sincerely believe that the new standard must succeed if the Internet is to grow, and assume that the transition is inevitable because of the presumed depletion of the IPv4 address resources. However, by examining the associated network effects, developing the economic parameters for transition, and modeling the underlying economic forces which impact network operator decisions, the study paints a more complex, nuanced picture and reaches several important conclusions.
The good news for IPv6 is:
- IPv6 won’t become an orphan. Major content providers and fast-growing mobile Internet services are deploying IPv6 and gradually diverting traffic away from the old Internet to native IPv6 networks. It appears as if IPv6 will dominate some parts of the newer, larger-scale parts of the Internet.
- For network operators that need to grow, particularly mobile networks where the software and hardware ecosystem is mostly converted, IPv6 deployment can make economic sense. It mitigates a major constraint on growth and reduces the operational complexities and costs of large-scale Network Address Translation.
But there is also bad news for IPv6:
- The incentives of large, fast-growing network operators aren’t the same as those of many other networks that make up the Internet. Many enterprise networks don’t need to grow much and/or may still be lodged in a slower-moving software and hardware ecosystem tied to IPv4.
- Diffusion is still in early days. Of the 215 economies measured, only 26 (12%) had steadily increasing levels of IPv6 capability over the three year study period. Another 18 countries (8%) had measurable levels of IPv6 deployment but exhibited plateaus in adoption, with IPv6 capability stalled for several years at levels anywhere between 8% (Austria) and 59% (Belgium). This group included numerous mature European economies as well as Canada and Australia. A large majority of economies (169, or 79%) had no appreciable IPv6 deployment. Countries in this group were located in all regions, and included small (Mozambique) and large (China) economies.
- Because of the costs of IPv6 deployment, there is a strong correlation between higher GDP and IPv6 deployment levels.
- Networks that deploy IPv6 must maintain backwards compatibility with non-deployers. This imposes a cost penalty on IPv6 users and eliminates some network effects that would degrade or cut off networks that do not convert.
- Even if they have deployed IPv6, growing networks must continue to acquire scarce, increasingly expensive IPv4 addresses to interconnect with the rest of the Internet. Deploying IPv6 does not immediately end the problem of IPv4 address exhaustion.
The study examined the impact of IPv4 address markets and address scarcity on the transition. It points out that cloud service providers, who need to serve both Internets, are the buyers of the largest amount of IPv4 numbers.
The report concludes that legacy IPv4 will coexist with IPv6 indefinitely. A variety of conversion technologies, and more efficient use of IPv4 addresses using NAT, will support a “mixed world” of the two standards for the foreseeable future.
The study can be freely downloaded in our Research Section.
An online internet governance system built on a common global framework could accelerate closing the IPv6 gap.
I am not sure how a “common global framework” (whatever that means is dependent upon ones point of view) could affect the interoperability issues that exist between the two versions of IP.
The simple matter is that the original intent was for SIPP (now known as IPv6) was supposed to be fully compatible with the DoD Internet Protocol Standard (now known as IPv4) but alas, it is not. This single “error” is the reason for why we could not just extend the network to new networks and hosts.
Therefore until such time as someone can create a “magic” solution, every network operator must still use the existing IP protocol and worse, needs additional IP space to add new features or add additional customers.
So Internet Governance plays no role in forcing this transition.
There is little one can do to sprinkle “magic” onto fundamentally incompatible addresses. Why aren’t IPX and IP “compatible”? Because they *aren’t*, plain and simple; they are different layer three protocols. However, the layering principle does give us the best magic, viz, that all the meaningful application support is in fact compatible between these two network layer protocols. Unlike with IPX, for example, with IPv[46] the exact same socket semantics work in both protocols, indeed when correctly coded there is literally *nothing* else an application must do to support IPv6 along with legacy IP. UDP and TCP are literally the same, and most other relevant IP protocols also have direct compatibility (or are irrelevant).
To go beyond dual stack hosts to support single stack IPv6 requires some notion of compatibility at the network layer. Some kind of NAT is needed for this “compatibility”, and that is NAT64 (of course with DNS64). Once again, this pretty much “just works”, and, unlike traditional legacy NAT, is needed less and less as more and more of the Internet adjusts to the current generation of the Internet Protocol. NAT64/DNS64 is considerably simpler to deploy than legacy NAT — just define your /96 prefix and away you go. There are a few corner cases where the legacy IP address is actually embedded into the application or similar NAT-insensitive artifacts, but as a result (ironically) of legacy NAT, this is very uncommon. NAT64 provides the needed translation and is extremely simple to deploy. With this, hosts can safely operate as single stack IPv6.
At this point, the cost of deploying and supporting IPv6 is vanishingly small compared to the total costs of operating a network.