What's in a Name? - New Challenges for DNS

A Brief History of DNS

Back in the days of the ARPANET, all systems had addresses containing system.arpa, and the mapping between those names and the addresses was kept up-to-date in a host table maintained by Stanford Research Institute Network Information Center (SRINIC). As the Internet grew, it became clear that the system would not scale. A new system was created that offered a tree-branch hierarchy of domain names, with all domains emanating from a series of seven top-level domains (TLDs): .edu, .com, .org, .gov, .mil, .net, and .int plus country domains. The system was originally developed in 1983 by the DNS team of Paul Mockapetris, Craig Partridge, and Jon Postel (see interview with Jon Postel on page 37 and profile of Paul Mockapetris on page 18). It was officially adopted by the major network providers in January 1986. The most recent document that describes this structure is RFC 1591 by Jon Postel [Postel94].

DNS As Resource Allocator

DNS became useful for more than just mapping names to addresses. It became a way to alias systems and to direct electronic mail and other services to alternate locations. It also acts as a rudimentary filter or security mechanism through the use of inverse addresses. When you connect to a server, the server can ask what IP address you're coming from and look up the address, so that there's some level of telling where you're logged on from. This usage is weak, however, because you can't rely on everyone's having an inverse name or the same inverse name each time the person uses the network.

Some sites, such as www. yahoo.com, are extremely popular on the Internet, and a feature called round-robin DNS is used so that when the nameservers are queried, they give back one of several addresses for www.yahoo. com. Thus, www.yahoo.com's load can be distributed over several servers, but users need remember only one address. More on DNS as a load-balancing mechanism can be found in [Brisco95].

Work continues in the DNS arena, with special emphasis on mobile computing and dynamic updates for DNS [Vixie96], so that regardless of which IP address you come in on, you will have the same host name. In addition, dynamic updates will become commonplace, with renumbering [Carpenter96] and with dynamic address allocation under things like IPv6. With this we will undoubtedly see DNS becoming a more central Internet technology.

DNS As a Social Identity

What's in a name? Everything. Domain names have become more than a network address; they have become powerful marketing tools because of their ability to easily identify-and sometimes locate-a company's or organization's presence on the Internet. Given the growing ubiquity of Internet usage and its potential for mass adoption, it is understandable that companies, organizations, and individuals feel strong about their domain names. The value of Web and e-mail addresses such as toothpaste.com and perfume.com is becoming equal to that of a logo. A few years ago it was impossible to imagine that anything as unintelligible as http://www.abcdefg.com would appear anyplace other than a programmer's handbook, let alone a television advertisement or a billboard. Now it's even hard to sit in a restaurant without overhearing exchanges of e-mail and Web addresses. It makes sense that businesses want to be sure they have mycompany.com under their belts, just as many companies in the United States have done with 800-MY-COMPANY.

Domain names have also formed a rudimentary directory service. A well-known commercial company that's on the Internet is probably in the .com domain, so it's fairly easy to find companies such as ibm.com, dec.com, sun.com, and others. On the face of it, searching for companies in the .com domain makes sense, though it is interesting to note that the DNS system was never intended to serve in that role.

One of the limitations of the .com domain space is that there can be only one xyz.com. When two companies have the same identity or trademark, only one can have the xyz.com domain name. Even if there is only one trademark for that name, someone else may have come along and taken the domain, for anyone can apply for a domain name that is available. Lawyers and policy analysts, as well as representatives from both large and small companies, have turned up the volume on discussions surrounding this issue. Compounding the problem is that trademarks are often a local issue and the Internet is worldwide. How then do companies protect their trademarks and their identities if the two can be used as domain names by others?

Additional such dubious practices are cropping up in the Domain Name System arena, like the registering of trademarked names by one other than the trademark holder, who then resells the domain to the holder or some other interested party. Are domain names becoming a marketable commodity? Should policies be put in place to prevent this kind of practice? If so, by whom?

There is little protection for trademarks with respect to domain names outside the standard legal recourse for trademark disputes.

At this time, there is no registry watchdog group for your trademark to help arbitrate domain requests in order to determine whether one infringes on your trademark. This has led to some companies' registering trademarks as domain names, even if those domain names will not be the principally used domains of the company.

Different registries have dif-ferent policies with respect to handling trademark disputes. Some will simply leave it to the normal legal process for resolution; others have a dispute resolution policy requiring provision of proof of registration of the trademark. A survey of dispute resolution policies is available from Digidem at http://www.digidem.com/legal/domain.html.

Several notable lawsuits related to trademarks and domain names are pending, but unfortunately there hasn't been clear legal precedent set to determine how trademarks are to be treated in cyberspace and how the registries are to respond to trademark issues. Such legal precedent or bodies of case law would clarify the role of the registry and either give responsibilities to the registry or specify the way the parties must resolve via standard legal recourses.

Just Who's in Charge Here Anyhow?

The Internet Assigned Numbers Authority (IANA) is currently responsible for the Domain Name System and has delegated the operational Internet Registry to the InterNIC. For more information on how the IANA fits in with the other Internet bodies, consult [Hovey96].

Once the IANA decides which top-level domains exist and how they're delegated, the TLDs are carried by a set of root DNS servers that serve as the base for the DNS directory on the Internet. The reality of how the Internet functions is that the Internet service providers ISPs-who maintain the connections-cooperate and by joint agreement use the central root servers designated by the IANA.

Here are how some of the delegations are currently handled:

· .edu, .com, .org., and .net are administered by the InterNIC, which also handles registry for North America.

· .gov is delegated to the U.S. Federal Networking Council and is administered at the InterNIC.

· .mil is managed by the U.S. Defense Data Network.

· RIPE NCC handles registry for Europe, delegating most of the country TLDs to national registries.

· APNIC handles registry for the Asia-Pacific region, delegating most of the country TLDs to national registries.

When the Internet was still funded by the U.S. Department of Defense, it was clear who made the decisions. In today's Internet, things run because several organizations and entities have agreed to cooperate. There exists, however, a centralization of registries that has led to growing dissatisfaction within the Internet community; that is, the InterNIC retains a virtual monopoly on managing the .com domain. Several proposals have been put forth, and efforts are currently under way to change that situation to provide more options for people looking for domain names and to make the registries more competitive.

Many of the efforts revolve around new TLDs, which in itself won't necessarily alleviate the trademark problem, because some companies may attempt to register their trademark in every new TLD. The issue probably won't be settled until a legal precedent has been set in the courts.

Earlier this year Jon Postel authored a proposal-with input from many people-on the establishment of new top-level domains [Postel96]. Under his plan, new registries would be granted licenses to run TLDs on a competitive basis, with an ad hoc group-made up of representatives from the IANA, the IETF, and the Internet Society-to oversee the application and approval process. ISOC would form the legal umbrella to cover the IANA in the effort.

The new registries would have their own new TLDs, but there is the option that in the future, TLDs would be shared by more than one registry. For example, .com addresses would be available from more than one registry. Postel's proposal is believed to be the one that has the most support of Internet community at large and of such groups as the Internet Society Board of Trustees.

Those who wish to learn how to set up their own DNS servers and participate in the Internet DNS structure might want to read the book DNS and BIND, by Paul Al-bitz and Cricket Liu (Sebastapol, Calif.: O'Reilly & Associates, 1992), and "Common DNS Operational and Configuration Errors," by D. Barr (RFC 1912, Feb. 1996).

Another proposal, by Simon Higgs, maps new commercial TLDs to categories of trademarks under the International Trademark Schedule of Goods and Services, which classifies trademarks by product or service sector [Higgs96]. Although it doesn't settle the issue of who serves as a registry, it does attempt to address the case in which two companies in different industries have the same trademark that they want to use in their domain name.

One approach that has been taken is to create a new set of root DNS servers that will handle both the current TLDs as well as newly created ones. The AlterNIC provides a way of registering new TLDs in their root servers to create an alternate domain universe consisting of current TLDs and new ones created by AlterNIC. The new TLDs work only for those who use the alternate root DNS servers. Currently, this approach has experienced only limited acceptance by the Internet community and leaves open the question of what to do if other groups start setting up their own alternate root name servers.

A number of workshops have been held so far: Internet Names, Numbers and Beyond: Issues in the Coordination, Privitization and Internationalization of the Internet, which was held November 20, 1995, in Washington, DC, and sponsored by NSF and the Harvard Information Infrastructure Project (IIP); Internet Administration and Infrastructure, held in February 1996 in Washington, D.C., and sponsored by the NSF and the Internet Society; and Coordination and Administration of the Internet, held September 8–10, 1996, and sponsored by Harvard IIP, CIX, ITU, and the Internet Society.

The broader issues surrounding DNS have been subjects of discussion by the IETF, the Internet Architecture Board, and the ISOC Advisory Council, among other arenas. No firm consensus has taken place, but conversations continue. Clearly, the issues affect users, Internet service providers, DNS registries, and the various Internet standards bodies.

An article written by Robert Shaw of the ITU offers much in-depth information on these issues, links to the workshops, and related documents and is definitely recommended reading [Shaw96].

One of the most active mailing lists discussing DNS and new TLDs is the newdom mailing list, which you can subscribe to by sending e-mail with the subject "subscribe newdom" to newdom-request@iiia.org.

Conclusions

DNS has grown up to be much more than just a way of mapping names to IP addresses, and it will continue to evolve. It provides a focal point for both technical and political discussions covering how the Internet is run-and by whom.

DNS discussions and proposals taking place now will help set the stage for future direction of the Internet. It's an important topic that affects all users, administrators, ISPs, and policy makers, and it requires due diligence and care in addressing these issues to keep the Internet healthy.

Acknowledgments

I'd like to thank the following people for their reviews of this article and helpful comments: Brian Carpenter, CERN; Tom Newell, InterNIC; Robert Shaw, ITU; and Alex Tognino, IBM. Of course, all the mistakes are mine.

References

Note: Internet Drafts and RFCs are available from the InterNIC at http://ds.internic.net/ds/dspg0intdoc.html.

[Brisco95] Brisco, T. "DNS Support for Load Balancing" (RFC 1794, Apr. 1995).

[Carpenter96] Carpenter, Brian, and Yakov Rekhter. "Renumbering Needs Work" (RFC 1900, Feb. 1996).

[Higgs96] Higgs, Simon. "Top Level Domain Classification and Categorization" (Internet Draft, draft-higgs-tld-cat-02.txt, Jul. 1996).

[Hovey96] Hovey, Richard, and Scott Bradner. "The Organizations Involved in the IETF Standards Process" (Internet Draft, draft-ietf-poised95-ietf-orgs-03.txt, Jun. 1996).

[Postel94] Postel, Jon. "Domain Name System Structure and Delegation" (RFC 1591, Mar. 1994).

[Postel96] Postel, Jon. "New Registries and the Delegation of International Top Level Domains" (Internet Draft, draft-postel-iana-itld-admin-01.txt, Jun. 1996).

[Shaw96] Shaw, Robert. "Internet Domain Names: Whose Domain Is This?" 1996 (available at http://www.itu.int/intreg/dns.html).

[Vixie96] Vixie, Paul, Susan Thomson, Yakov Rekhter, and Jim Bound. "Dynamic Updates in the Domain Name System (DNS UPDATE)" (Internet Draft, draft-ietf-dnsind-dynDNS-09.txt, Mar. 1996).