Alphabet Soup: The History of DNS
1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 

by Ross Wm. Rader

Domain Name Service acts as the backbone to the Internet. Like a fast, furious and omnipresent Vanna White, DNS is the key to the numbers behind the names and vice versa. It is the overarching entity that allows the Internet to function, a product of grand ideology and technical wizardry. And politics. Lots of politics.

Writing the history of DNS is akin to writing a comprehensive American history, or the history of the English language. Variables galore, involved parties pushing this way and that — some acting out of a desire for progress, others out of self-interest. Philosophies clash, government talks and monopolists pull strings. Sadly, like most protocols, the history of DNS is about 10 percent technical and 90 percent political. As is usually the case, no one tends to realize this until it's too late to make change. And the average person, the one generally the most affected, has little or no opportunity to participate in the process.

The one certain piece of the puzzle is that all involved parties realize the significance of owning even a piece of DNS — Vanna's the prize, and eager hands from all sectors are reaching for her bosom. This story will look into the dynamics that shaped the DNS system. Why is ICANN important to everyone who ever logs on to the Internet? What role does the United States government actually play behind the scenes? How the hell did Network Solutions find its way into the catbird's seat?

The answers to these questions have, and will continue to, determine what the Internet looks like today and what it will look like tomorrow. It's a long, strange and sometimes convoluted tale, and it's one worth knowing. As George Santayana said: "Those who cannot remember the past are condemned to repeat it."

The Early Years

"In [the early days] there was no 'commercial motivation'..."

— Dr. David Mills

The evolution of DNS has been long governed by technicians and politicians — and often the same people assumed both roles — making it difficult at times to look back at events that critically shaped the protocol and determine exactly what happened. Anthony Rutkowski, vice president of Internet strategy for Network Solutions Inc., states, "The evolution of the Internet's method for names and number assignment is one of the Internet's most enduring and difficult transitions."

When Doug Engelbart created the ARPAnet NIC (Network Information Center) at the Stanford Research Institute in 1967, DNS did not exist. The network was small enough that the users and servers generally knew how to get around from service to service and interact with each other without the benefit of a global directory structure. As the network grew in size, it became apparent that such a service would be important, but due to scale reasons it was never quite taken to its conclusion. Instead, in 1971, Peggy Karp conceived of "host mnemonics," or more simply, Internet names.

The Internet Request for Comments (or RFC) documents are the written definitions of the protocols and policies of the Internet. Building on the concepts contained in RFC 226, Karp created a lookup table that mapped all of the network resources in one text-formatted file. Called HOSTS.TXT, the table contained all of the hostnames and their related IP addresses. Operators would install this file on their local server, which would then gain the capability to perform the requisite lookups locally and enable the computer to find resources out on the larger network without a lot of overhead. Whenever an operator added a new machine to the network, he or she would complete an email template with the appropriate information and send it off to the appropriate people at Stanford Research Institute. They would then compile all of the changes, include them in the next release of HOSTS.TXT and store the new file on a globally available FTP server. Operators would retrieve the updated versions on a regular basis and install them on their local servers. The first version of this table was distributed in 1972. This arrangement worked well for a number of years, but it suffered from one systemic problem — it wasn't scalable. Ultimately, ARPAnet's success was the lookup service's undoing, and engineers concluded that a new structure would have to replace HOSTS.TXT.

DNS, Baby

The Domain Name System was conceived in 1981 by Dr. David Mills, a COMSAT engineer at the time. In RFC 799, he outlined the concepts and facilities required for an Internet Name Domains system that would eventually scale to facilitate addressing of "thousands of hosts." Although a working solution to the much larger problem resulted, Mills' initial intent was a quick fix.

"My interests [in creating Internet Name Domains] were more focused on the mechanics of doing this and on mail forwarding principles for the Internet," Mills recently said. "Not the least of my concerns were the mechanisms for handing off mail between forwarders and handling errors as they might develop."

The solution would ultimately address a problem far larger than what Mills had originally foreseen. As Mills observed then, "it will not be practicable for every Internet host to include all Internet hosts in its name-address tables. Even now, with over 400 names and nicknames in the combined tables, this has become awkward." The current DNS supports millions of host entries.

RFC 819 was cowritten the following year by Jon Postel from the Information Sciences Institute (ISI) at the University of Southern California and Zaw-Sing Su from Stanford Research Institute. It built on the earlier work by Mills, and gave the first general outline of the DNS structure and how it would allow for easier cross-network access. In November 1983, Dr. Paul Mockapetris, also from ISI, wrote "Domain Names — Concepts and Facilities" as RFC 882 and "Domain Names — Implemen-tation and Specification" as RFC 883. These two papers outlined a completely new way of managing host name lookups. Most importantly, they included two very important concepts: authority and delegation.

Domains, like sri.com, have authority over the sub-domains that fall beneath them: mail.sri.com. These are called zones; authority exists on a zone-to-zone basis. Delegation describes the process by which someone gains authority over his or her zone. SRI Network Operations might delegate authority to their Hong Kong office for hk.sri.com. The Hong Kong office would then have authority over the hk.sri.com domain and all sub-domains below it — for instance, mail.hk.sri.com and so on. The Hong Kong office could also delegate authority to a third party for the fourth-level domain.

Delegation, as it stands, is the basis upon which all power and conflict in the DNS today is derived.

When the design work was finally complete, Jon Postel and Joyce Reynolds wrote RFC 920, which outlined the groundwork that had to take place and set the rapid pace at which the change would occur. Their work also was significant for another reason: It outlined the initial top-level domain names that would be added to the DNS when it was finally deployed. These included .com, .net, .org, .edu, .gov, .mil and .arpa. The document also planned for the creation of country-related top-level domains (TLDs) using the two-letter International Standards Organization (ISO) codes — becoming .ca, .us, .uk, etc. Top-level domains were also planned for "multiorganizations" — groups that were "large, and composed of other organizations; particularly if the multiorganization can not be easily classified into one of the categories and is international in scope." The concept of multiorganizational TLDs was later assumed under .int — an international top-level domain.

The successful deployment of DNS made it apparent to the Defense Communications Agency (the DCA, now called the Defense Information Systems Agency or DISA) that a centralized management structure was necessary to manage the root and delegate authority to registrants within that root. As the DCA had split the ARPAnet into the ARPAnet and MILnet in 1993 — with MILnet becoming integrated into the Defense Data Network (DDN) shortly thereafter — it was important that whoever they chose to manage the technical operations of the root and the registry did so in a manner consistent with the requirements of the military.

The DCA chose the Stanford Research Institute to manage the registration of all domain names in the new DNS. It awarded to ISI the root management and operations of the Defense Data Network Information Center (DDN-NIC), a semi-private digital packet switched network meeting the United States Department of Defense's stringent communications standards. After all, who better to manage the services than those who had created it? In March 1985, all of the work paid off and the first domain names were registered — some claim the very first .com was symbolics.com; others claim it was think.com. By most accounts, 1985 was the year that the "Modern History" of DNS actually began.

For a number of years, the evolution of DNS progressed smoothly and functioned very much as back-office support to the operation of the network. In 1986, the growing needs of the academic community led to the creation of the NSFnet by the United States National Science Foundation, which awarded the backbone management contract to Merit Networks Inc. in 1987. The NSFnet was the last incarnation of the network prior to the Internet that we use today. By the time that the ARPAnet was retired in 1990, the network of networks had grown to include more than 100,000 connected host computers.

The demise of the ARPAnet was certainly the end of the grand experiment and proved that the concept of a globally distributed, fault-tolerant, standards-based network was indeed a practical one. It was the first step toward the commercialized environment that we know today. It was also the last time that politics was incidental to the DNS. Very quickly, politics would completely overshadow what the wizards behind ARPAnet had set out to accomplish.

Toward Commercialization

"What we call 'Progress' is the exchange of one nuisance for another nuisance."

— Henry Havelock Ellis

In 1991, the Defense Information Systems Agency awarded a small contract that would change the face of DNS. The contract specified the terms under which a new third party would take over the administration and maintenance of DDN-NIC, which had been up until this point under the management of Stanford Research Institute. A defense contractor, Government Systems Inc. (GSI) was awarded the bid in May. By late September, GSI had assumed operational responsibility for DDN-NIC.

Although the official record indicates that the contract was fulfilled by GSI internally, GSI actually outsourced it to a small private-sector contractor, Network Solutions Inc. As Rutkowski points out, a number of people fell under the impression that NSI was a spin-off of GSI, due to their similar names and the fact that NSI was performing the entire DISA-NIC contract. But that was not the case.

Leveraging the experience gained in managing the DDN-NIC operations for DISA, Network Solutions tendered a bid in May 1992 with the National Science Foundation to provide network information services management to NSFnet and NASA Research & Education Network (NREN). NSI was awarded this contract in October of that year under the terms of NSF Cooperative Agreement # NCR-9218742 which, among other things, specified "The Non-military internet registration services to be provided under this agreement will initially include, but not be limited to, the following: Domain name registration, Domain name server registration, Network number assignment, Autonomous system number assignment."

This was an enormous development in DNS history, and its importance was two-fold. First, the academics who had built the service were now no longer involved. Second, the central nervous system of the Internet was squarely in the hands of the private sector. Certainly NSF still exerted a great deal of control over the activities of Network Solutions and made sure that the agreement was executed in a manner consistent with the needs of the network and its community. But the seeds of privatization had been sown.

1994 was a watershed year in the development of the Internet. First, it was the year that saw the beginning of the popular adoption of Tim Berners-Lee's World Wide Web with the release of NCSA Mosaic. Second, the National Science Foundation and the federal government started to remove themselves from the picture, allowing commercial interests to start exploiting the network. Suddenly, everyone wanted to get online, surf the net and dive into Gopher-space. The rush toward a connected society had begun.

Until 1995, academic policy on the name-space allowed anyone who had access to a name server to register a domain name with the NSI InterNIC — for free. Needless to say, as the perceived value of being online increased, so did the perceived value of Internet domain names. Thus, warehousing and speculation was born. It wasn't unusual for a speculator to register hundreds or even thousands of domain names simply on the possibility that someone might want to purchase the domain name in the future.

In an attempt to mitigate this, as well as recover some of the costs of operation, NSF began to allow NSI to start charging $50 per domain name registration this same year. Needless to say, this did not please the general Internet public.

The NSF was eventually forced to drop 30 percent of the $50 fee that had been earmarked by the U.S. government for general Internet infrastructure projects. The infrastructure fund was deemed an illegal tax because Congress had not approved it. The NSF as a governmental body had no mandate to levy fees of this nature on the populace. During the short time period that this fee was collected, hundreds of thousands of dollars were collected outside of U.S. jurisdiction.

"The DNS Wars"

"The Internet is about consensus, not truth. Never mistake truth for consensus."

— Brian Reid

Network Solutions' decision to charge for domain names started a period in DNS history that some refer to as the "Domain Wars." Battle lines were drawn around the thorny issue of who ultimately controlled the namespace in fact, and who should control the namespace. These opposing camps could be roughly characterized as The Old Guard, The Geeks, The Monopolists and The Politicos. Unfortunately, it was seldom easy to identify who belonged to which camp. Many people held multiple allegiances. Each of these players saw an opportunity to leverage control of the namespace into something much larger — what that something was varied from group to group based on its agenda. The rallying point for these groups often centered on Network Solutions' role as a monopoly within the DNS. Partially answering this question, Jon Postel released the first of his international Top-Level Domain drafts in May 1996. These drafts were based on prior work of Larry Landweber, which had been revised and iterated upon by Randy Bush, Karl Denninger and Brian Carpenter, each pioneers in their own right. Postel's first draft put forth a number of very important concepts that completely define the political structure of DNS today. It also sparked a continuing unresolved debate.

Postel's goals were very simple. His draft outlined a mechanism to allow open competition in domain name registration within the namespace, thus muting the monopoly that NSI had. It also provided a way to give the Internet Assigned Numbers Authority (IANA) the legal and financial umbrella it needed to survive. Later that month, the board of directors of the Internet Society (ISOC) voted to accept Postel's draft as a proposal for the management and expansion of the namespace at their annual meeting in Montreal, Canada. While the board recognized that significant work was needed to complete the plan, it was thought that it could be achieved in a matter of months. This meeting would ultimately lead to the creation what would become known as the IAHC, or Internet Ad Hoc Committee.

Founded in October 1996, the IAHC aimed to "undertake defining, investigating, and resolving issues resulting from current international debate over a proposal to establish global registries and additional international Top Level Domain names (iTLDs)," according to Don Heath, who was then president and CEO of the Internet Society. The founding board read like a who's who of the inside track of the Internet, with some interesting new faces as well. The International Telecommunications Union (ITU), the World Intellectual Property Organization (WIPO), the International Trademark Association (INTA), ISOC, IANA and the Internet Architecture Board (IAB) all appointed members to the IAHC board. The IAHC quickly became symbolic of the efforts of The Old Guard to change the face of the namespace, and it quickly announced to the world it would create seven new top-level domain names — .firm, .store, .web, .arts, .rec, .info and .nom. The IAHC also called for the creation of a competing group of domain registrars. It issued a report concerning the administration and management of generic Top-Level Domains (gTLDs). The plan — known as the gTLD-MoU, or Generic Top-Level Domain Memorandum of Understanding — was released to the public on Feb. 4, 1997. It expanded on the concepts originally put forth by Postel and wrapped them in a number of compromises designed to satisfy the many interests involved in the process. The trademark community was represented, as were the privacy advocates, standards bodies, and Internet and telecommunications carriers. These groups were joined to become the Interim Policy Oversight Committee (IPOC), a bootstrap committee that would later cede control to a permanent Policy Oversight Committee. Further, the IAHC's report described in detail a new not-for-profit registry model that would be managed by a consortium of international registrars.

Some viewed the gTLD-MoU as simply a more complex and unworkable version of the earlier efforts of IAHC, Postel and ISOC. Others thought it was a scary step toward a new world government led by the ITU and WIPO. These concerns, and many more, consumed almost all of 1997 in debate. The international flavor of the arrangements was a concern to many. The registry would be run by a Swiss non-profit (CORE Internet Council of Registrars); dispute resolution would be managed by the WIPO, a Swiss-based United Nations treaty organization; activities would be supported by another Swiss-based organization, the ITU.

While the IAHC Old Guard was pushing the gTLD-MoU as a done deal, the dissident voices were starting to shout. Some of these voices coalesced into an organization known as eDNS. eDNS was a loosely knit group of individuals who came together to create an "alternative root namespace." Led chiefly by Karl Denninger and Eugene Kashpureff, eDNS attempted to replicate what Network Solutions and IANA had already accomplished with .com, .net and .org. eDNS also set the pace for The Geeks involved in the process. Their efforts proved to be short-lived. Infighting between the two leaders eventually led to Denninger disassociating himself from eDNS, which he characterized as "a disgusting joke begotten from what it was envisioned to be" in his resignation letter.

While Denninger remained active around the fringes of the process, Kashpureff escalated his efforts to dethrone Network Solutions. In July, he altered the DNS for InterNIC.net, which was then Network Solutions' primary face on the web, so that all traffic flowed instead to his new effort, AlterNIC — not once but twice. AlterNIC was essentially a protest vehicle for Kashpureff's view of the NSI monopoly. In a statement to CNET, Kashpureff said, "The hack was a result of a year's worth of work … I'm proud of what I did." However, he conceded: "I have some remorse. I understand that what I did was wrong. My ideals, my emotions got the best of me." Kashpureff ended up fleeing the country for Canada, where he was eventually jailed for three months pending extradition. Upon his return to the United States, his penalties were a $100 fine and probation.

While The Old Guard debated and The Geeks revolted, another incident took place which, had it occurred at a different time, would have been nothing more than a temporary anomaly. On July 17, 1997, human error at Network Solutions corrupted the master zone files for .com, .net and .org, throwing the Internet into a state of disarray. While the problem was resolved in roughly four hours, the repercussions of the meltdown were felt by the Internet community throughout the rest of that day and into the next. Not only did this event occur the same week that Kashpureff "hijacked" the InterNIC, the story was broken by The New York Times' John Markoff, a prominent writer known for his coverage of and subsequent book about Kevin Mitnick. With the story making headlines in the Times and many other major news outlets, people quickly realized that the Internet wasn't as indestructible as they had been led to believe. In fact, many realized that the stability of the Internet rested largely with Network Solutions.

Through this uncertainty, President Bill Clinton directed the Secretary of Com-merce to "privatize, increase competition in, and promote international participation in the domain name system" as part of his administration's Framework for Global Electronic Commerce. Accord-ingly, the Department of Commerce issued its first request for comments on July 2, soliciting "public input on issues relating to the overall framework of the DNS system, the creation of new top-level domains, policies for registrars and trademark issues." This was widely viewed as a very important step. For the first time in the history of the Internet, the government of the United States had proven it was aware of such arcane subjects as DNS. It was also the first indication that the deal The Old Guard was pushing was perhaps not quite done.

This confluence of events and uncertainty spurred the debate from which the American Registry for Internet Numbers (ARIN) grew. ARIN is a non-profit organization established for the purpose of administration and registration of Internet Protocol (IP) numbers for North America, South America, the Caribbean and sub-Saharan Africa. Until 1997, InterNIC/NSI had taken care of these responsibilities, but as Kim Hubbard, past-president of ARIN puts it, "After NSI began charging for DNS registration and the discussion on DNS registration competition began, there was concern voiced among some in the industry that IP registration might somehow be affected."

Others in the business have a decidedly different view. "ARIN was created as a non-profit entity that Network Solutions could more or less spin off because there was just not going to be any money in that part of their business," said a source close to the issue who declined to be named for this article. Regardless of the primary motivation, it was quite widely held that the community had to move forward with the ARIN idea.

"DNS registration competition discussions at one point became quite hostile and the last thing anyone wanted was IP registration to be pulled into the political morass that DNS had become," said Hubbard. "There was concern that while it was plausible to make DNS registration a financially competitive function, the same could not be said for IP registration. For various technical reasons, it was not viable for companies to compete on a monetary basis to allocate IP numbers."

Once the community had recognized the potential problems, ARIN was created quickly. "It became evident that the best way to handle the situation was to separate the two registration processes in every way possible," Hubbard said. Through work with NSI, the Internet community and in consultation with the Federal Networking Council, ARIN officially opened for operation on Dec. 22, 1997.

The situation with CORE and the gTLD-MoU was not progressing any more smoothly at this point. The House of Rep-resentatives held hearings on the state of the DNS that were specifically concerned with the activities arising from the original IAHC plans. Opposition to the gTLD-MoU was at an all-time high as original supporters dropped out. Further, the involvement of the government gave the opposition a chance to derail the entire process that was still being sold as a "done deal" between the various parties and the Department of Commerce. Success appeared to be at hand. But on Sept. 30, 1997, the deal start to unravel.

At one of the many House Committee hearings on the subject, opponents of the IAHC/gTLD-MoU plan seemed to gain the upper hand. The Commercial Internet Exchange (CIX), Information Technology Association of America (ITAA), Tony Rutkowski representing World Internetworking Alliance (WIA) and Andy Sernovitz representing the Association for Interactive Media (AIM) all went on the record in some manner opposing the plan. Perhaps most damaging to the gTLD-MoU was Sernovitz's denunciation of IAHC and IANA as betraying the United States to the governments of Libya and Iraq. While the claims were never completely substantiated or formally investigated, they damaged the credibility of the group behind the gTLD-MoU. Further, the total of all the comments successfully managed to call into focus the potential shift of power that the Internet represented to a country other than the United States.

Even though much damage had been done at the hearings and The Old Guard was now largely on the defensive, the gTLD-MoU continued to press forward. In October, CORE, awarded the technical management contract to run the registry system needed to support the new top-level domains and the shared registry system to Emergent Corp., a small U.S. consultancy.

1998 — A Make-or-Break Year

On Jan. 30, 1998, Ira Magaziner, who was the senior adviser to President Clinton for policy development, released a discussion paper that later became known as "The Green Paper." According to Jonathan Weinberg, who was working close to the matter at the time, the Green Paper was the result of thinking within the higher levels of the government that it could better run the Internet. And, it appeared that Magaziner had the authority of the President to make certain that U.S. rule of the DNS was maintained.

While Magaziner's Green Paper was very similar to Postel's original iTLD drafts, he made one very important distinction. The United States government would remain involved over the short- and mid-terms to ensure a reasonable and orderly transition to a new, nonprofit organization, NewCo. This statement would effectively unravel ISOC's position as the driver behind the IAHC/gTLD-MoU process. It also spoke out directly against the trademark resolution processes that WIPO had written into the gTLD-MoU. While Magaziner was in favor of trademark protection as part of NewCo's mandate, he made it very clear that the domain registrars, not NewCo, should hold the cost of dispute resolution and all liability.

Magaziner also outlined the early demise of the Network Solutions agreement with the federal government. His major points on this subject follow:

Magaziner's Green Paper specifically indicated that NSI's monopoly was scheduled to end later that year, on Sept. 30. For the first time, there was a clear indication from the highest levels that the government-endowed monopoly that NSI enjoyed was finite. The IAHC/gTLD-MoU detractors suddenly had a unified proposal to rally around, The Old Guard was left with a plan in shambles, Network Solutions was faced with a severe limitation of power and profits, and Jon Postel was not happy with the new direction.

In response to the Green Paper, Postel issued this statement: "I am in agreement with the main theme of the proposal ... I am less comfortable with the details of the proposal on how new generic top-level domains, registrars and registries would be established, and the restriction to only five new gTLDs."

On the weekend preceding Feb. 4, Postel undertook what he characterized as a "transition plan test." During this test, he had the root DNS system reconfigured to obtain DNS information not from the A root server operated by Network Solutions, but rather by one that he operated at ISI. "I wanted to see how easily management of the root servers could be passed to another machine when the government gives up its control of the domain system as described in the Green Paper," Postel wrote at the time. The government did not react well to this test. According to a report published in Network World, Becky Burr, a senior official with the Department of Commerce, said the government knew nothing of the test beforehand. "The timing is unfortunate," she said.

Although control of the DNS was quickly handed back to Network Solutions, and therefore the government, Postel had made it very clear that he was unhappy with the statements made in the Green Paper and that he was still very much in charge of the DNS — regardless of who had contractual control.

On Feb. 17, unknown individuals broke into Best Communications (now owned by Verio) and stole the registry servers that Emergent was installing for CORE. While this event was most likely the result of unfortunate timing, many at the time attributed the theft to a conspiracy backed by Network Solutions to discredit CORE and IAHC. Others seemed to think that this event was nothing more than an insurance fraud designed to underwrite the mounting costs faced by CORE. To this day, it has not been established who was responsible for the break-in or what happened to the equipment, but it was clear that CORE now had some serious obstacles to overcome.

On June 5, Magaziner released a revised version of the Green Paper that took into account the extensive public commentary that had been collected by the government since the release of the original draft. This new document — referred to as "The White Paper" — was released as an official statement of policy on behalf of the Department of Commerce. Most relevant was Magaziner's call to establish consensus on the issues presented in the White Paper by the relevant stakeholders. Essentially, he was granting the Internet pioneers one last chance to determine what their fate under the White Paper would look like.

Following the release of this document, a number of groups scurried to organize and forge toward consensus. The most important of these emerging groups was the International Forum for the White Paper (IFWP). The IFWP genesis can be traced back to the announcement of a slightly earlier group, the Global Incorporation Alliance Workshop (GIAW) that had come out of nowhere and announced a conference to be held in Washington July 1-2. The prevailing sentiment quickly emerged that the GIAW was nothing more than a Network Solutions front to address the issues presented by the White Paper in a manner amenable to NSI shareholders. As Rick Wesson of Alice's Registry Tools noted at the time on an IAHC mailing list, "It would seem NSI and all the folks that lay claim to the goldmine of TLDs are setting up a meeting ... the scary part is it looks like only those folks that oppose the gTLD-MoU are coming." Minutes from an organizers' conference call show that there was agreement on a need to "diffuse some of the negative associations with the GIAW." Thus, the IFWP was born.

The IFWP described itself as "an ad hoc coalition of professional, trade and educational associations representing a diversity of Internet stakeholder groups." The group opened its meetings to all Internet stakeholders and encouraged them to support the ongoing process.

The IFWP gathered support for a few very important reasons. First, it was quickly gaining momentum at a time when the IAHC/gTLD-MoU/CORE process was losing steam. Second, each group that had felt left out of the IAHC process was specifically embraced by the IFWP. Lastly, Network Solutions' support of the IFWP lent it the credibility the IAHC lacked.

The IFWP sponsored a number of regional workshops throughout the world in an attempt to address the issues raised by the White Paper and define a specific course of action that could fulfill the terms put forth by the government. While the IFWP, it appears, did not produce a full working draft of bylaws for NewCo as requested by the White Paper, it succeeded by becoming a catalyst for the process the White Paper contemplated.

The Birth of ICANN

Over that summer, Postel wrote a new draft set of bylaws for a "new IANA" that would satisfy the role set forth by the White Paper. Postel positioned the new IANA with "responsibilities in three interrelated areas: Internet Protocol addresses, domain names, and protocol parameters." The responsibilities included the root server system, as well as the work then carried out by the existing IANA. The new IANA's goal was to "preserve the central coordinating functions of the global Internet for the public good." Postel released these drafts throughout June, July and August, and the most important was released on Sept. 17, with the benefit of the endorsement of Gabe Battista on behalf of Network Solutions. The goal of these new drafts, wrote Postel in September, was to "capture the best ideas from all sources, including the International Forum for the White Paper, the business community, the Internet technical community and other stakeholders." That draft is known to the community as IANA/NSI Draft 4. Draft 5 followed a few weeks later on Sept. 28 — curiously without the explicit support of Network Solutions.

Other groups, finding the NSI/IANA drafts lacking, engaged in a similar effort in an attempt to define NewCo. Most notably, The Boston Working Group was concerned that the NSI/IANA drafts lacked a clear set of checks and balances to ensure democracy within the new organization. Further, they posited, "the IANA/NSI proposal is neither a product of the IFWP process nor does it conform to the IFWP consensus points. Nor does it meet the standards required by the NTIA White Paper, either in terms of content or the process through which that proposal was derived."

On Oct. 1, the NTIA announced that NewCo (or the new IANA) would be called the Internet Corporation for Assigned Names and Numbers, or ICANN. This same day, the cooperative agreement between the U.S. government and Network Solutions had been scheduled to terminate, and the transition to ICANN would start. Postel was now mainly concerned with ensuring that the transition to ICANN would happen on terms that were agreeable and fair to everyone involved.

A week later, Joe Sims, Postel's lawyer, appeared before the House of Representatives Subcommittee on Basic Research and the Subcommittee on Technology of the Committee on Sciences. Postel, having recently taken ill, was unable to testify in person, though his testimony included a statement describing his feelings: "Most of the work to come will be done by others; IANA will continue its technical work and I will of course stay involved in the process, but it is time for ICANN to begin to lead this effort." His words, although simple, foreshadowed far more than anyone could imagine. On Oct. 16, 1998, Jon Postel passed away.

The Feeling-Out Process

With Postel no longer involved, the process wavered somewhat. However, with most of the groundwork laid by the flurry of drafts produced earlier that year, others were able to pick up where he left off. Through the remainder of October and most of November, Esther Dyson, Joe Sims, Mike Roberts and others worked towards satisfying the comments issued by the NTIA on the IANA drafts. On Nov. 25, the Department of Commerce announced that it had finalized negotiations with NewCo and officially recognized it as the Internet Corporation for Assigned Names and Numbers for the purpose of transitioning DNS management from the government to industry.

Gabe Battista resigned as CEO of Network Solutions mid-November, and Ira Magaziner left at the end of December. It appeared that there were very few people left standing outside of the new ICANN.

Although ICANN had made it through the Domain Wars and was deemed worthy enough to assume the mantle from the Department of Commerce, the organization still had some significant challenges ahead. Now, it was actually faced with overseeing the DNS, bringing competition to the namespace and adding new gTLDs to the root in an open, transparent and bottom-up manner. With an appointed interim board of directors in place, ICANN tackled the thorny issue of competition first. ICANN chose to pursue implementation of a Shared Registry Service with Verisign that would see Network Solutions split off into two entities, the registry and the registrar, while new competitive registrars would be accredited. The threat originally raised by Magaziner had come to pass.

House Trumps Nsi's Trump Card

Just before the Department of Commerce officially recognized ICANN as NewCo/New IANA, Network Solutions played its trump card. Jim Rutt, the company's new CEO, entered into a revised cooperative agreement with the Department of Commerce — the original agreement had given NSI its monopoly; it had been regularly amended afterward. Amendment 11 of the agreement specified that Network Solutions must work with NewCo through the transition specified by the Department of Commerce. Rutt entered into the agreement on Oct. 7, 1998.

As such, if NSI refused to recognize ICANN as being the NewCo described in the cooperative agreement, ICANN would have no one with whom to enter into an agreement. This put ICANN in a tough spot — but it nevertheless pushed forward with its plans to introduce competition. Network Solutions' tactics delayed the effort significantly, as the only way that ICANN could get NSI to do anything was to first file the request with the Department of Commerce, which would then relay it to NSI. Despite the odds against it, ICANN announced on April 25, 1999 that it had selected 34 companies that would be accredited to compete with Network Solutions for the registration of domain names.

Five of these companies would be allowed to participate in a special test-bed that had been designed to allow a limited number of companies to work out the technical issues associated with the new Shared Registry System that NSI was implementing. On June 7, register.com Inc. announced that it had successfully registered the first domain name under the new competitive regime. The test-bed, originally scheduled to last until June 24, was extended until Sept. 10, and then Nov. 30.

NSI continued to refuse to recognize ICANN through the summer of 1999. The tactics worked reasonably well, until the situation came under the scrutiny of a House Commerce Committee investigation into ICANN. The panel, chaired by Tom Bliley (R-Va.), grilled Rutt. One of the most widely reported exchanges occurred between Jim Rutt and House Representative Bart Stupak (D-Mi.).

Stupak: NSI claims that it only has to recognize ICANN if ICANN has a "final agreement." DoC says the agreement has been finalized. Sounds like a delay tactic to me. What do you think? Have you ever told ICANN or DoC that there is no final agreement?

Rutt: Let me ask my lawyer and get back to you tomorrow. Yes, I have said there is no final agreement.

The press jumped on the story as hard as the committee had jumped on Rutt. As ABCNews.com reported, "Rutt told the committee that although his company controls about 75 percent of the names on the World Wide Web and the Internet, it does not have a monopoly — which drew snickers from the standing-room-only crowd. Democrats on the committee were even more skeptical of NSI's claims that ICANN was out to destroy its business."

Network Solutions' stalling tactics were now out in the open, and the Commerce Committee did not take them lightly. Industry insiders widely regard the July hearings as Rutt's Waterloo. Nonetheless, Network Solutions still maintained enough leverage with ICANN and the Department of Commerce to push the issue closer to the concessions that NSI wanted.

Between Nov. 2 and 4, 1999, ICANN's first-annual general meeting was held in Los Angeles. The agenda was specifically geared towards ratifying a proposed agreement between ICANN and Network Solutions. Everyone agreed a truce was necessary, but few felt it could be pulled off. While ICANN kept a firm grip on the proceedings, there was a new sense of empowerment among certain groups, a feeling that anything could be accomplished given the right effort. There was an unspoken understanding that this agreement would make or break ICANN. If the parties did not successfully arrive at a conclusion on this issue, it was almost certain that the structure would disintegrate.

The original agreements tabled provided Network Solutions with significant concessions. Through a number of informal meetings held by a number of interests through the week, various positions were readied for presentation during the Nov. 3 public comment forum. The comment forum was especially raucous. People were lining up for hours to present their views. In some cases, such as with the registrars group, dozens of proposed revisions to the agreement were put forward to the ICANN Board. Somehow, the board of directors and ICANN staff managed to reconcile most of these presentations and to negotiate a revised agreement with Network Solutions through the night. On Nov. 4, ICANN and Network Solutions agreed to a truce. Among other things, the agreements provided specifically that Network Solutions would have to separate its registry and registrar businesses; that it would receive an extension to its operating agreement if it divested one or the other business in two years time; and most importantly, that it recognized ICANN as NewCo. As expected, a few weeks later, the Department of Commerce also accepted these revised agreements.

Nearly 15 years full of political and ideological angling after the first .com was officially registered, today's system of domain registry was finally in place.

The Here and Now

The ensuing year went relatively quietly. The introduction of new registrars precipitated an average price drop from $35 per year to roughly $15. Registration rates rose rapidly. ICANN occupied itself with the introduction of new generic top-level domains. Although the introduction process took quite some time, the ground covered was well trod by the previous debates. On Nov. 16, 2000, the ICANN board of directors finally approved the introduction of seven new top-level domains: .aero, .biz, .coop, .info, .museum, .name, and .pro. While this process is not yet completed, it is expected that the general public will be able to register these domain names sometime during the last half of 2001.

ICANN's work is not yet done, nor have the dynamics of DNS become any less complex. The industry is still struggling with internationalizing the namespace, the impact of private root systems and a myriad of other issues. ICANN itself is also still the center of a maelstrom of controversy, but, for better or worse, work is being done and progress is being made.

And that, in a nutshell, is a complete history of DNS. Take a deep breath and take it all in, because the final outcome is still a work in progress. We'll keep you posted.


Author's Note:

All the material for this article was gathered through hours of research on the Internet, including web sites, mailing list archives and the venerable IETF RFC server. While every attempt has been made to verify that the actual events occurred as documented, incomplete transcripts, poor memories and non-existent web archives made this a difficult task. If you feel that an important event was missed or unfairly portrayed in this article, please drop me a line. Thanks go out to everyone who took time to contribute, including Anthony Rutkowski, David Mills, Tim Denton, Richard Sexton, Jonathan Weinberg, Antony Van Couvering and many others. Special thanks go out to the countless thousands who built the system and implemented the policy that made it worthwhile to write this article in the first place.



 Alphabet Soup: The History of DNS 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 



Print version of: http://www.whmag.com/content/0601/dns/