IPv6: Time's still not right

It's been more than 10 years since the development of IPv6, yet it's had virtually zero adoption among U.S. enterprise customers and service providers, even though the U.S. Office of Management and Budget recently announced it intends to require support across government agencies for IPv6 by 2008.

Are network managers and service providers sticks-in-the-mud who refuse to get with next-generation technology?

Not at all. The dirty little secret behind IPv6 is that although it's touted as "next-generation" IP, purportedly increasing security and QoS, in reality it adds little to existing IP specs. The security and QoS capabilities built into IPv6 are virtually identical to those added over the years to IPv4.

To continue reading, register here and become an Insider. You'll get free access to premium content from CIO, Computerworld, CSO, InfoWorld, and Network World. See more Insider content or sign in.

It's been more than 10 years since the development of IPv6, yet it's had virtually zero adoption among U.S. enterprise customers and service providers, even though the U.S. Office of Management and Budget recently announced it intends to require support across government agencies for IPv6 by 2008.

Are network managers and service providers sticks-in-the-mud who refuse to get with next-generation technology?

Not at all. The dirty little secret behind IPv6 is that although it's touted as "next-generation" IP, purportedly increasing security and QoS, in reality it adds little to existing IP specs. The security and QoS capabilities built into IPv6 are virtually identical to those added over the years to IPv4.

All that IPv6 really does is increase the number of directly addressable Internet endpoints (to about 340 trillion addresses). This is potentially useful, particularly in a world in which every individual soda can, let alone every vending machine, might conceivably need an IP address.

However, this feature isn't exactly free. Quadrupling the address space dramatically increases the bandwidth required to transport each packet. Sending a 64-byte message, for instance, requires 250% more bandwidth in IPv6 than in IPv4.

Obviously, the overhead increase is greatest for small packets, which make up a minority of the data transferred across today's Internet. But it's still a non-negligible issue given that one of the imagined drivers for IPv6 is the notion of vast networks of tiny sensors at the end of presumably very-low-bandwidth links.

There's the issue of the appropriate next-generation routing architecture. Most folks assume the IPv6 routing will be a simple "scaling up" of today's routing architectures, but some point out that the next-generation Internet will require next-generation routing. And that routing has yet to be envisioned, let alone implemented.

Finally, there's the real question of why a network manager would want directly addressable endpoints. Most companies run firewalls with network address translation (NAT) precisely to cloak their endpoints from the Internet. Moreover, using NATs dramatically increases the number of addresses available to networks behind the NAT device, which eliminates the one clear driver for IPv6.

Where does the momentum behind IPv6 come from? Two places: First, large organizations and service providers outside the U.S. have been forced by a lack of addresses to adopt IPv6 (when IP addresses were originally allocated, these folks got too few). Second, as noted, the U.S. government has mandated its use.

It remains to be seen how rapidly the deployment will take off. Government mandates have a mixed track record in driving civilian technology deployment (anybody remember GOSIP? How about ISDN?). And it's clear from the government's announcement that they have a limited understanding of how difficult the transition is likely to be, which means the time frame is likely to be extended well beyond 2008.

At some point, IPv6 support may be required to connect with non-U.S. service providers and government offices. But until that happens, no need to switch.