Stop the IPv4 World, I Wanna Get Off

The transition from IPv4 to IPv6 on a global scale is a colossal chore. The task is so extensive, that several major ISP vendors have asked everyone to hold their horses, at least until a smooth trail can be blazed.

by Jim Thompson [April 5, 2002]

The Internet Engineering Task Force (IETF) wants development on new IPv6 transition tools stopped—at least for the time being. But this doesn't mean that the updated addressing protocol has been laid to rest permanently. IPv6 is still being pushed forward by developers in Asia and Europe right now. But IPv6 won't receive the same kind of love and attention from U.S. developers for another year or two.

Controversy and confusion reigned late last month when the IETF pronounced at their March meeting in Minneapolis that it's time to bring some sanity to the rampant development of IPv6 transition tools. Their solution—stop all development of new tools in order to nurture the move from IPv4 to IPv6.

"We did a hard reset and decided that there will be no new development until we describe a set of environments that we believe are common," said Tony Hain, co-chair of the IPv6 Transition Working Group, otherwise known as Next Generation Transition (ngtrans). "This will give us the chance to describe the particular tool set that should be used for a particular environment."

Some developers see the decision as a blow against the open creativity, a direct slap in the face for progressive developers who are working on new applications and tools to smooth the transition period. Hain cautions against jumping to conclusions concerning the request to take transition tool on a brief hiatus.

"The IETF is not saying that new tools aren't useful and needed, but that we don't need new things right now. We have to decide the best use for what we have now, before moving into new areas," noted Hain.

Brief hiatus
The call for a temporary halt in new IPv6 transition tools development is expected to take only a few months. The process involves re-evaluating where IPv6 technology is at today, determining which existing transition tools should be used tomorrow, and eventually agreeing about what needs to be done to resolve any lingering technical issues.

The good news is that the protocol has reached the stage where it is stable enough for this grand transition process to even be considered. After all, a half-baked protocol transition tool would simply not stand up to the scrutiny the IETF intends to put it through. The majority of work over the next few months will focus on clarifying existing documentation and deciding on the best method to address known issues.

One of these issues is determining best practices for finding domain name servers. Using auto discovery mechanisms for addressing means that Dynamic Host Configuration Protocol (DHCP) is not in use.

Dynamic addressing, as used in IPv4, simplifies network administration because the software, not an administrator, keeps track of IP addresses. Consequently, adding a new computer can be done without having to manually assign unique IP addresses.

The question is—how do you find a DNS server without DHCP? The answer is that the IETF needs to determine best practices before IPv6 can be effectively deployed.

Another issue is how best to transition the world from IPv4 to IPv6. Hain said that six years ago, he pushed to have dual IP stacks placed in all nodes so a host could use either protocol, or fallback to IPv4, if IPv6 were to fail to connect.

Is everybody happy?
This is a good idea, but it's not practical for every node in every country. "Particularly in Asia and Europe there were complaints because they don't have enough IPv4 addresses, so they are trying to move as quickly as possible to IPv6 without IPv4," Hain said.

But the pressure to move directly to IPv6 brings with it the potential of severing network ties from Asia and Europe with the rest of the world. Those using only the limited address space available under IPv4 cannot see the networks operating only on IPv6, since IPv4 only allows for 32-bit address space, and namespace under IPv6 is extended to 128-bits.

From the beginning of this debate, the primary reason for moving to IPv6 is the need for additional address space. In Asia, where getting addresses is nearly impossible, developers are anxious to switch to the new protocol. This sense of urgency is exemplified by the Japanese government, which declared that switching to IPv6 a "national priority" and mandated that government, educational and industrial networks in Japan achieve IPv6 deployment as soon as possible.

Spurred by a shortage of IP addresses, Europe is also moving quickly to implement the IPv6 standard. However, back in the U.S. address space is far more plentiful, so the dot-com leader doesn't share the same sense of urgency. U.S. resistance to make the move to IPv6 is especially strong within the enterprise space where a massive infrastructure of network management and tools makes switching to IPv6 both painful and expensive.

Several industry leaders, including Cisco Systems, Microsoft and Compaq have been encouraging both individuals and enterprises to embrace IPv6 for several years. But encouragement from major vendors has not produced sweeping IPv6 implementation among major network operators.

Hain believes the demand for IPv6 implementation will come from consumers, at least those using gaming consoles and PDAs. Mobile device demands will also be a major force behind promoting the new protocol, but not in the immediate future. "When the mobile space is ready, there will be a big play to hurry [IPv6 implementation], but mobile won't be first," Hain said.

According to Hain, the first serious push to make IPv6 the worldwide addressing standard will begin in Asia later this year and continue on into early 2003. Additionally, Europe will begin their drive in earnest around next year. But the U.S. won't get serious about IPv6 uptake until 2005. "There is a clear indication that by the end of the decade [2010] the party will be over for IPv4," declared Hain.

—End