Section 10.10. Cost of Introduction

10.10. Cost of Introduction

Some people believe that the introduction of IPv6 is way too expensive, and they don't even start to think about it. Other people just want to know what it will cost. The next section contains aspects to consider and some thoughts on the business case for IPv6.

10.10.1. Hardware and Operating Systems

When planning ahead for IPv6, the cost for hardware upgrades is not extensive. In many cases, IPv6 is not part of the hardware and can be installed as part of the operating system or as a software upgrade. Most operating systems ship with IPv6 included at no extra cost as of a couple of years ago. If IP functions are implemented purely in hardware or if a system's software cannot be upgraded, the hardware must be replaced to support IPv6. If you plan ahead, though, this can usually be accomplished with the next upgrade in the regular lifecycle of the product and therefore generates no additional cost. As our case studies have shown, this is confirmed by organizations that have deployed IPv6.

10.10.2. Software

The time and effort needed to port applications to run over IPv6 depends on how the application is written and how it accesses the IP layer (if at all). If the application strictly separates the application layer from the communication layer, the porting is simple and quick. If the application uses complex middleware and customized APIs to communicate with the IP layer, the time and effort needed for porting is in direct relation to the complexity of the middleware. In this case, all dependencies on IP must be modified to support IPv6 (address length, header information, etc.).

It is to be expected that common applications already support IPv6 or will do so in their next major upgrade version. In this case, you can limit the cost for software by planning early and using the regular lifecycles of applications.

All proprietary and self-developed applications need to be tested for conformance and analyzed to find the best way to port them, or to find out whether they are too complex to be ported and need a rewrite to become IP independent. A simple porting of an application makes sure it runs equally well with IPv6 transport. A creative porting of an application may include using the advanced features of IPv6 and thereby extending the flexibility and functionality of the application. This can even be seen as cost associated with introducing advanced applications based on new technologies more than as cost of introducing IPv6.

10.10.3. Education

Every technology upgrade requires education: for developers, vendors, service providers, and the infrastructure and systems operators in organizations. For home users, it should be their ISP's job to make the transition simple.

A well-planned education program according to job responsibilities supports a smooth introduction of IPv6 substantially. For system operators, the time and effort needed for the learning is not higher than for maintaining the IPv4 infrastructure. We are used to integrating new technologies all the time to keep our networks state-of-the-art. We had to introduce DHCP, NAT, and VPNs in the past, and we mastered them. Now we are going to introduce IPv6 and the challenge is no bigger. And it is worth effort, time, and money because in the long term, the maintenance of an IPv6 network will be less costly than the maintenance of an IPv4 network. People that have a good understanding of IP and have mastered the previously mentioned technologies will have no problems mastering IPv6.

It is well worth it to take some time to learn about IPv6 before going into production. There are new concepts and possibilities in IPv6, and we need some time to become familiar with these, learn how to best make use of them, and then integrate the things we've learned into the planning.

10.10.4. Planning

The most important aspect of integration is the planning. The same rule applies here. Extending an infrastructure to keep it state-of-the-art always requires planning and should be seen as an investment. Instead of planning the next extension of IPv4, we will now plan the integration of IPv6.

The planning requires network architects and systems engineers with a good understanding of IPv4 and networking concepts. The first thing they need is a thorough IPv6 education and an IPv6 playground. The planning should not focus on how to make the same services available over IPv6. It should include understanding the new features of IPv6 and making use of them to create new concepts of architecture, security, mobility, and administration.

The introduction of IPv6 is smoother and cheaper the earlier you plan. A step-by-step integration is the best way to go, as it gives you time to integrate what you learn as you go. Step-by-step integration is possible thanks to the many available transition mechanisms.

10.10.5. Other Costs

Probably the highest costs arise when you wait too long. The longer you wait, the more investments you make into maintaining and extending your IPv4 infrastructure. This is money and effort you invest in an old technology. You may need to build band-aids for IPv4 (e.g., NAT) that later complicate the introduction of IPv6.

There are no reasons to tear down a performing network that fulfills all the requirements, but as soon as you have to invest in fixing or extending your IPv4 infrastructure, you should stop for a moment and consider IPv6 as an alternative. As mentioned earlier, putting IPv6 as an evaluation criterion on all your IT shopping lists for products that have more than two years' lifetime is a good idea. You may not want to turn IPv6 on tomorrow, but you may want to do so next year. If your equipment and software are ready, you can do it without additional cost when the right moment comes.

Another cost factor can be if a business-critical application comes out that is based on the advanced functionality of IPv6. If your infrastructure is IPv6-ready, you can introduce that application with moderate cost. If at the same time you have to master the transition to IPv6, this might create significant cost and put your current infrastructure at risk (plus create some headaches and sleepless nights).