IPv6 – Picturing the Transition

Paul Wilson, Director General, APNIC

IPv6 is the next generation of IP, the Internet Protocol. Due to its very large address space, IPv6 will allow the Internet to continue to grow, as a single global end-to-end network, to thousands or millions of times its current size. Without it however, Internet growth has nearly reached its limit, due to the limited size of IPv4 address space. In this article, Paul Wilson, Director General of APNIC, answers some important questions about the pending transition to IPv6, which should help the non-technical Internet user to come to grips with this crucial topic

Download article as PDF

Paul Wilson joined APNIC in 1998 as the Director General. As part of this role, he represents the activities and interests of the Asia Pacific Internet Community in local and global forums related to the development and management of the Internet.

Prior to joining APNIC, Paul accumulated ten years technical and business experience in the Internet industry, including consultation on various Internet projects for the United Nations and other international agencies.

In 1989, Paul, as Technical Director, helped set up Pegasus Networks, the first independent ISP established in Australia. From 1992 to 1997, he was the Chief Executive officer, and oversaw the successful growth of the company as a renowned service provider in Australia. During this time, he was involved with the establishment of the Association for Progressive Communications (APC), and served for several years as an APC Council and Executive Board member.

During the 1990’s, Paul also consulted on various Internet projects for the United Nations and other international agencies, including the International Development Research Centre (IDRC). From 1994, he worked as a principle consultant on IDRC’s Pan Asia Networking (PAN) Program, a program aimed at introducing and developing Internet services in developing economies of the region. In that capacity he worked on projects in many locations including Mongolia, Vietnam, Cambodia, Maldives, Nepal, Bhutan, PNG, and China; while also working on similar activities in several countries of Latin America.

The promotion of IPv6 has been underway for most of the past 10 years, but a widespread understanding of this important issue, and its implications, has been slow to emerge. There are many questions about the transition to IPv6: Why is it needed at all? What will I have to do? What about compatibility? What will be the cost, and the benefit? At APNIC, the Regional Internet Registry for the Asia Pacific region, we have been working on this issue, and on helping to answer these questions, for much of the past decade.

Why is the transition necessary?

The Internet is a global network requiring globally-unique addresses for all of its components, and it is growing rapidly. For many years in fact, the Internet has outgrown its existing address supply, to the extent that many networks use a technique of address sharing (known as “network address translation”, or NAT) that can support only a subset of available services to their users. Running and expanding the Internet without a supply of additional addresses is like trying to grow the telephone network without more telephone numbers; in both cases the available solutions are generally unsatisfactory.

The intended solution for the Internet is to transition to a new version of the IP protocol, the fundamental communications standard of the Internet, that supports a far larger number of addresses. This new standard is known as IPv6. However, the Internet is far too big to allow us to upgrade the whole network at one time: the cost and logistics of such an exercise are simply prohibitive, and there can be no “flag-day” for IPv6 implementation. Instead we are facing, necessarily, a distributed upgrade process that is commonly referred to as the “IPv6 Transition”. The IPv6 transition will be carried out across the Internet, and across individual networks, as a prolonged process. Some parts of the Internet will keep running IPv4 for some time, and, in time, some will be built with only IPv6; but in the majority of cases, the component networks of the Internet will run both IPv4 and IPv6 at the same time for an extended period of time. This latter configuration is known as a “dual-stack” architecture, because it involves both IP protocol “stacks” operating at the same time.

So the “IPv6 transition” is not an event, as if IPv6 would simply replace IPv4. It is important to stress that IPv4 and IPv6 will coexist on the Internet, and in many networks and devices, for many years to come. IPv4 may in fact never disappear entirely from the Internet, but for our practical purposes we could agree that the transition will be complete when IPv4 is generally considered “optional” for users and service providers. That time is still at least 5 years away, and probably closer to 10.

What about Compatibility?

Much has been made of the fact that IPv4 and IPv6 are different and “incompatible” protocols, and questions raised about the technical choices made in IPv6. Let’s look at the way such technical decisions are made.

In any complex system, the process of introducing change has many considerations. These include not just a question of cost and benefit, but also practicality, business models, service continuity, investment lifetimes and public policies. In engineered systems, particularly those relying on standards of some kind, “backward compatibility” is a critical question – because such a concept allows a new technology to be introduced gradually, while avoiding to some degree at least the cost of discarding old technologies. In many cases, a series of new technologies may be introduced which are backward-compatible in a useful way.

A recent historical example is in the evolution of vinyl audio recordings through various standards: mono, stereo, quadraphonic; 78, 45 and 33RPM. As time went by users could upgrade their equipment and still play their old vinyl records, which was clearly important. But at some point a major change was needed, namely the transition to CDs, and in that case a harder decision had to be made, involving a more expensive upgrade path and much wider effect.

In the case of the Internet, we have reached not only the limits of IPv4 addresses, but also of the various mechanisms available to extend its life. Therefore we are at the critical point where a more significant change is needed, as we were in the evolution from analogue vinyl to digital CD recordings. In likewise in the IPv6 case, a backward-compatible extension is simply not feasible; instead we need to provide interoperation as an additional function, through translators or “dual stack” networks. In reality, the translator approach raises a significant set of security and technical issues, while the “dual stack” approach offers the best path in terms of gradual transition while maintaining security and service integrity.

But in any case, and similarly to LPs and CDs, IPv4 and IPv6 are in fact highly “compatible”. They both run on the same wires and fibres, and they can run through the same devices; they both carry the same Internet applications, with minimal recoding; and they perform the same functions. Indeed the capabilities offered by these two protocols are so similar that most users should not see any change in the way in which their connected device behaves. At the level of network applications, the transition is intended to be entirely seamless and invisible, while of course the details and complexities are hidden, and handled by the various service providers.

At APNIC we have drawn another analogy, between the IPv4-IPv6 transition and the transition of our transportation system from gasoline to electricity. In the case of cars, you certainly don’t try to fill you electric car with petrol, or vice versa; but the two vehicles run on the same roads and obey the same rules; they function in the same way, so that a driver needs little re-education; and they provide the same services, so that passengers actually don’t need to know the difference.

What will happen and what will it cost?

The answer to this question is that it depends entirely on who you are, what your relationship is with, and your level of dependence on the Internet.

Under the transportation analogy, let’s ask the same question: “What do I need to do?” The answer clearly depends on your relationship with, and your dependence upon, fuel in transportation. If you are a driver, you need to know where to get your fuel, and that’s about all. If you run a gas station, you will want to think about how to provide the new fuel to your customers. If you run a company fleet, or a mechanics shop, or an accessory business, then you will have other things to think about; other threats and other opportunities. So it is with IPv6.

As a customer of a successful and well-run Internet Service Provider, you should not be aware of the introduction of IPv6 at all, in terms of your services or costs. You will, ideally, find IPv6 supplied to you transparently, supported by your latest upgraded DSL router, working automatically and invisibly.

As a customer of another ISP however, you may find that IPv6 has not been considered at all, and that after some time your IPv4-only connection starts to deliver erratic service, for at least some applications and services. Or, that you cannot get as good performance in accessing those services as others do.

As a website owner, you should know that your online customers are your bread and butter; and that delivering information to them is your primary mission. As new networks are built with IPv6, more Internet users will have IPv6 as their primary protocol, with IPv4 only used as a secondary mechanism (and in many cases, through translators which impede their connectivity). So if you want to maximize your market, and the positive experience of your customers, you need to be sure that you provide your services on IPv4 to those still using it, and also on IPv6, to those new customers who must use it (that is, a “dual-stack” solution).

Another way to look at the transition is in terms of a customer-supplier model; for the Internet is really a vast network of not just electronics, but of service providers and customers. A website owner will have many customers of course, but it will also have many suppliers: an ISP providing upstream connectivity; a datacentre providing accommodation; a vendor providing the server itself; and a team of experts providing consulting, security, web development and backend services. Up and down this chain, Internet services are being provided in one form or another, and up and down this chain, IPv6 services must also be provided. For when you think “Internet” you need to think “IPv6”.

So no matter what your relationship is with the Internet, now is the time to Think IPv6. In most cases this is a simple matter of asking your service providers (be they ISPs, vendors, technicians, consultants or your own staff) what they are doing about IPv6. Certainly if you rely on the Internet for your survival, then this question is one of survival, so you need to understand and trust the answer you get. You may find that you need to take some action this year, or next year, or maybe this month; but in any case it is now time to understand that action, and when it needs to be taken.

For more information visit:
www.apnic.net