Every computer connected to the Internet is designated by a unique number known as an Internet protocol, or IP, address, which in a sense is akin to a telephone number. For example, when a user types "www.cnet.com" into his or her browser, the request is routed to the number 126.96.36.199. Domain names differ in that they are the translation of IP numbers, making Web addresses easier to remember.
The current system for assigning IP numbers, known as Internet protocol version 4 (IPv4), can supply 4 billion unique numbers. But only a small fraction of the numbers can actually be assigned to computers on the Net due to inherent inefficiencies in the system. Some engineers and policy makers warn that the Net's explosive growth may mean that IP numbers are exhausted in the next ten years.
"If China wants to put a billion people online, there's no way to do that no matter how you slice it [under the current system]," said Scott Bradner, a director at the Internet Engineering Task Force, the standards body that approved the new protocol. In addition, Bradner and many others envision the day when household appliances, telephones, and cars are connected to the Net. "That kind of network appliance thing eats addresses very quickly."
A new system, which has been on the drawing boards for nearly a decade, seeks to remedy the potential shortage as well as alleviate other problems. But critics say its costs to businesses far outweigh its benefits.
The new system, IPv6, recently passed two important milestones. Last month, for example, Internet pioneer Vint Cerf, now with MCI Worldcom, led formation of a coalition to back the system that includes technology heavyweights such as Sun Microsystems, Microsoft, and Cisco. A week later, the Internet Assigned Numbers Authority, which is responsible for allocating Internet addresses, issued numbers based on IPv6 for the first time.
IPv6's advantage, proponents say, is that it offers exponentially more address space, just as a license plate with eight digits offers many more unique numbers than one with four. Proponents say IPv6 adds other important improvements, such as better security and support for wireless phones and other network devices.
But those pushing adoption of the protocol face tremendous obstacles because of the expenses businesses would incur to make the change. It resembles the challenge a company faces when changing its physical address--it has to print new business cards, send out change of address notices, redirect its mail, change its phone numbers, and the like. With IPv6, businesses would have to make massive infrastructure changes to their networks, which are costly and complicated.
The dilemma facing proponents of IPv6 is not unlike what engineers faced five years ago when they tried to convince banks and other businesses that the Y2K problem warranted tremendous resources to fix. The problem seemed too remote and the solutions too complicated and expensive for most businesses to grasp at the time.
Last month, when the Internet Assigned Numbers Authority began issuing numbers based on IPv6, it said the event was a "historic moment in the continued development of the Internet." But critics don't share that sentiment.
"What I'm desperately trying to get the [Internet Engineering Task Force] to understand and face up to is that the solution they put forward is not going to be acceptable in the marketplace," said Noel Chiappa, a networking consultant and critic of IPv6. "The fact that [the Internet Assigned Numbers Authority] is now assigning v6 addresses [may be] a milestone for IPv6, but if IPv6 is destined not to catch on, it's obviously not an important milestone for the Net as a whole."
Chiappa's predictions seem to be confirmed by the American Registry for Internet Numbers, which distributes IP numbers in North and South America. To date, only one company has registered an IPv6 number, said Kim Hubbard, the group's president.
"Most of the focus [on IPv6] has been in the research and academic areas," Hubbard said. "No [Internet service providers] are getting involved yet."
Chiappa and other critics say that IPv6 fails to fix the Net's biggest routing problems, and it introduces new glitches. "It's just another full employment act for software engineers," said Paul Vixie, the architect of several programs that route traffic on the Internet.
Instead, critics say, engineers would be better off taking a slower approach and spending time developing workarounds to the current system. Such workarounds include so-called network address translators and dynamic addressing, which radically reduce demand for IP numbers. Those workarounds, however, lack some of IPv6's advantages, such as improved security and support for mobile devices.
Cerf, the honorary chairman of the IPv6 Forum, agreed that most businesses are "pretty near" apathetic about the problem and said his job is to lobby them to move over to IPv6. He added, however, that this isn't the first time engineers have had to go to bat for major changes to the Internet infrastructure.
"The course of Internet evolution is almost never smooth," Cerf said in an interview. "Someone has got to get out there and goad people and say, 'You've really got to pay attention to this.'" Cerf said he faced a similar challenge decades ago when what was known as the Arpanet became today's Internet, requiring network administrators to overhaul their systems.
Other proponents of IPv6 agree that the stakes are high for revamping the system now, while it's still stable. They say critics exaggerate the difficulty of moving over to IPv6 and predict that as Microsoft, Sun Microsystems, and other software companies incorporate the technology into their programs, many firms and end users will run it without even knowing it.
They also warn that continuing to add workarounds to IPv4 will only encumber a system that was designed decades ago and was never intended to be a worldwide network on which the new economy rested.
"Since then, the Internet has grown and matured, and like any living organism, it has developed some bad habits along the way," said Charles Perkins, a protocol engineer at Sun and a consultant to the Internet Engineering Task Force. "IPv6 is a way for us to learn from our mistakes."