draft-skala-dnsreform-00.txt M. Skala Ansuz November 29, 1996 Reforming the Domain Name Service draft-skala-dnsreform-00.txt Status of this Memo This document is an Internet-Draft. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts. Internet-Drafts are draft documents valid for a maximum of six months. Internet-Drafts may be updated, replaced, or obsoleted by other documents at any time. It is not appropriate to use Internet- Drafts as reference material or to cite them other than as a working draft or work in progress. To learn the current status of any Internet-Draft, please check the 1id-abstracts.txt listing contained in the Internet-Drafts Shadow Directories on ds.internic.net, nic.nordu.net, ftp.nisc.sri.com, or munnari.oz.au. Abstract Problems with the current Domain Name Service are described and a solution to them proposed. This proposal solves flatness problems by encouraging a deeper DNS with more degrees of freedom in naming. It solves trademark problems by requiring unambiguous definitions of who may register under any new top-level domains. The proposal calls for an eventual end to the registration of second-level domains under the current international top-level domains according to the current rules, since the popularity of such registrations makes any significant change to the system impossible. 1. Introduction This proposal is intended as a starting point for discussion by the IAHC. It presents new iTLDs from the point of view of a small non-corporate system. Voices in domain-related discussions seem to come from corporate interests desiring marketable names and/or a portion of the money raised by registration fees. This proposal provides an alternate view based on the technical requirements the DNS serves. The domain name service, despite its usefulness as a means of providing corporate identity to businesses, was originally intended only as a way of assigning unique names to systems. Any proposal for changes to it must recognize first and foremost the need for providing unique names to all systems in a structure that the network can support in the long term. If the DNS can also serve advertising and directory purposes, that is good, but such purposes cannot take precedence over unique and sustainable naming. The current domain name service was designed with an hierarchical structure intended to distribute authority appropriately and prevent overloading any one level of servers. Its failure to meet those goals in its current application is due not to misdesign but to misuse. This proposal shows a better use of the DNS with little need for expanding it at the top level, while still allowing for such expansion to meet the inevitable desires of the users. 2. Considerations motivating this proposal The Internet's original design goals are being strained due to the abuse of the net by commercial interests. A business entering the network may possibly need to register a name. From the business point of view, the name should be relevent to the business, preferably identical or similar to the name it already uses off the net. It should be marketable and memorable. The considerations relevent to the network community are somewhat different. Domain names, above all else, must be unique. Numeric IP addresses are inadequate in a world of classless interdomain routing and IP version 6, where a host's IP address may change frequently. Domain names are necessary. If they are not unique, they are useless and the Internet becomes useless with them. Furthermore, domain names must be hierarchical. In an exponentially-growing network, the linear space of a single-level naming system is inadequate. This fact was recognized in documents like RFC-921 over ten years ago. Slightly less obvious is the fact that any system with a limited number of levels cannot survive forever in the face of exponential growth; a move to simple two-level domains will not correct the crowding problem indefinitely. Exponential growth always outstrips polynomial growth. 2.1. The DNS is too flat Because of the huge number of organizations registering second-level domains under .com, the DNS has lost its hierarchical nature. This flatness means that the server for .com must handle many thousands of domains. The namespace for such domains, despite its huge theoretical size, is rapidly becoming depleted of desireable names. Since every name has many potential claimants, each name is assigned more or less arbitrarily to one organization and all others must make do with less preferable alternate names. Organizations find it attractive to register many unnecessary domains in order to prevent competitors from using them, to cement a claim on all possible variations of a trademark, or to support a bid for dominance in a particular field of business. Another source of .com domains is the creation of so-called "virtual" domains: domains which do not name a new group of hosts on the Internet but instead provide an alias for an existing host, in order to create a more marketable URL for World Wide Web pages belonging to an organization without it own network connection. 2.1.1. Flatness at the root is not a solution It has been suggested that the creation of many new international top-level domains would resolve the flatness problem. As long as the current iTLDs remain open to new registrations under the current rules, they will remain attractive to new registrants who wish to emulate the naming scheme of old registrants. The creation of huge numbers of top-level domains creates cache overflow problems which could lead to a huge increase in demand on root name servers not surmountable by proposals for increasing the numbers of available root servers. Flat domains are bad at any level, but they are even worse at the top than at the first level. 2.2. Global uniqueness and trademark problems Internet domain names must be unique. Users, however, use them as trademarks and trademarks are only unique within a limited scope. The Internet is global. As a result, more than one potential registrant may present with a legitimate claim on a given domain name, and the registry is caught in the middle of a dispute better made in the legal system. 2.2.1. Proposal to ensure uniqueness This document's proposal partially eliminates the trademark problems by requiring a new kind of uniqueness. Every domain name under a new iTLD must have one unique potential registrant. If that registrant chooses not to register it, no-one else will be permitted to do so. Possible trademark disputes not solved by that uniqueness requirement are almost entirely shifted to their rightful arbiters, namely the national governments controlling or delegating registration under country code domains. 2.3. Misuse of the DNS debate The ongoing debate on the implementation of top-level domains has lead some network users to abandon the legitimate standards process in favor of forming their own. Such efforts will become more and more common if the network community does not create practical solutions soon. Grassroots drives for domain name reform, though admirable in principle, tend in practice to be tainted by local desire for money and power. 3. Phasing out single-level iTLD registrations The current system of international top-level domains presents little more than a single level of naming. One level is not enough to accomodate the demand on the DNS. As long as that one level exists, however, it will be seen as preferable by new registrants. Thus, any hope to replace it must eliminate and/or make undesireable registration under the old, flat DNS conventions. This section presents several possible ways to phase out single-level registrations under the current international top-level domains. These measures need not all be used or used in exactly the order specified. Here is a suggested timeline for these measures: * As soon as new iTLDs become available: stop accepting second-level registrations under current iTLDs from organizations that already have one * 3 months after new iTLDs: stop accepting second-level registrations served by hosts that already serve one * 6 months after new iTLDs: stop accepting all second-level registrations under current iTLDs except for delegate registries * Also, starting 6 months after new iTLDs and at intervals of a year thereafter: double the registration fees for existing second-level domains under current iTLDs except for delegate registries 3.1. One domain per organization The current system allows an unlimited number of registrations to each registrant. A good way to slow down the exhaustion of current international top-level domains would be to stop allowing new second-level registrations under the current international top-level domains to registrants who already hold such a domain. This may seem harsh and/or difficult to enforce, but it would only go into effect after new international top-level domains became available, potentially allowing unlimited registration. The point is to permit some continuation of current practices while encouraging organizations to move to the new system. 3.2. One domain per server A more aggressive limitation would seek to reduce the namespace depletion created by "virtual" domains, particularly under .com. The registries serving current international top-level domains must stop accepting applications for second-level domains whose nameservers already serve other second-level domains under current iTLDs. This effectively prevents organizations without their own network connections from using a domain as an alias for a WWW or similar site housed on another organization's computer. Of course, business interests have many legitimate and non-legitimate reasons to desire their own domain names even when they do not have their own network nodes. Under this proposal, organizations wishing to have virtual domains would be encouraged to register under new international top-level domains whose charters permitted it. Large numbers of business interests and especially WWW sites moving to new iTLDs would help cement the new iTLDs in the public mind and render them acceptable for other purposes. 3.3. No more exclusive second-level domains When new international top-level domains are firmly established as the preferred destination for new registrants, registries for current international top-level domains can stop accepting all new registrations for second-level domains not to be used by open delegate registries. 3.4. Replacing obsolete names Turning current international top-level domains into a three-level system is complicated by the thousands of existing two-level domains. When two-level domain names are no longer issued, it could become a status symbol, particularly for corporations, to have a name in the older style. There is no nice way to eliminate this problem. The net is now too large to support a "Flag Day" sudden switch to a more modern naming scheme. A controlling authority that mandated such a change would simply be ignored by network users, and likely replaced by inappropriately organized ad-hoc groups. A better solution is to slowly render old names unusable while making it as easy as possible to upgrade to three-level names. After the registries for current international top-level domains stop accepting new two-level domains for individual organizations, they must start increasing the fees for old registrations. Doubling the fees every year would be an appropriate rate. At that rate, the current price of US$50/year for a second-level .com domain would increase to $100 a year after the cessation of new second-level registrations, $200 a year after that, and so on. Current holders of such domains would have plenty of time in which to upgrade. However, the increasing rate (greater than $1.6 million after 15 years) would compel them to do so eventually instead of hanging onto old "status symbol" domains indefinitely. It is expected that the Internet Society would make arrangements with its contractor for some of the funds generated by this scheme to be applied to furthering the common good of the network, for instance by creating more local peering connections to reduce the demand on long-haul intra-continental links. Second-level domains under current iTLD registries, if relinquished by former holders, would be held unused by the iTLD registry for a specified period (suggestion: six months). After that time they would become available to new delegate registries. Second-level domains under current iTLDs formerly held for exclusive use by an organization would never become available for exclusive use again even if the iTLD registry were still accepting exclusive second-level domains in general. This underlines the policy that exclusive second-level domains under current iTLDs are to be phased out. 4. Delegated registries The current international top-level domains are impossible to eliminate at this stage in the Internet's development. Since that is the case, it seems reasonable to recycle them as a home for commercial delegate registries desiring exclusive control of a second-level domain name. 4.1. New second-level domains for registries The Internet Society's contractor controlling .com, .org, and similar domains would be responsible for evaluating applications from commercial registries desiring second-level domains under those international top-level domains. Guidelines for such evaluation could be drawn from the plans currently proposed for exclusively owned top-level domains; here are some thoughts on the kinds of criteria applicable: * Registries must not accept applications from organizations related to the registry; this is necessary to put to rest any suggestion of corporations running fake registries in order to get a desirable second-level .com or similar domain for themselves. * Registries must have defined procedures for dealing with disputes, and demonstrate that they have adequate funds to defend themselves against any lawsuits that may arise from such disputes. This is necessary to help prevent their customers from being left domainless by a registry collapsing in a legal dispute. * Registries must maintain appropriate server and communications resources to serve the domains they sell. * The name "www" would not be a permitted at the third level of delegated second-level domains under current iTLDs. * Registries would not be permitted to operate their own network connections at the second level of their domain (for instance, with email addresses of "name@registry.com"). They would be required to use one or more third-level domains in their delegation, or follow the same procedures as any other registrant for getting a domain elsewhere. This and the previous restriction are needed to keep delegate registries from unfairly having DNS "status" no longer permitted to other organizations. Registries capable of meeting these requirements would each be delegated one second-level domain under a current international top-level domain such as .com, and would be permitted to make whatever profit they could by selling third-level delegations beneath it. The Internet Society, or their contractor, would be responsible for checking to see that thes restrictions on delegate registries were being obeyed. Funding for such checks would be provided in part by the increasing fees for non-registry second-level domains under current international top-level domains. 4.2. Old second-level domains for registries Holders of current second-level domains under current iTLDs would be encouraged to register other organizations under their domain. However, they would not enjoy the fee status of new registries and so would be subject to the escalating fees described in section 3.4. of this document. They also would not be permitted a new second-level domain for delegate registry use as long as they held an old second-level domain. A holder of such a domain wishing to return it and apply for status as the delegate registry for that domain could do so, but would have to face the waiting period and possibly dispute the name with other potential delegate registries desiring the same domain. 5. Criteria for the creation of new top-level domains Any organization proposing a new international top-level domain must submit a proposal showing how the new domain fills a need not served by currently existing domains. The proposal must include a charter similar to those exemplified above, though likely much longer and more specific, defining the one entity eligible to register for each second-level domain. A standing comittee, preferably created by consensus of the Internet community but appointed by the ITU if necessary, must be created to evaluate the proposals. It need not be a large committee; the restrictions placed on new iTLDs by this proposal will make frivolous creation of them unattractive enough that only genuinely necessary international top-level domains will be proposed. This proposal does not specify the exact composition and procedures of the comittee, nor detail the technical requirements or fee structure applying to iTLD proposals, since the considerations governing those things will almost certainly change a great deal before the present debate is over. This document is a starting point. It is worth noting that under the following criteria the currently existing international top-level domains would be considered unacceptable. The proposed changes in section 3. of this document alleviate that inconsistency. 5.1. Unambiguous ownership Every new international top-level domain must have a clearly specified way of determining a unique possible registrant for every permitted second-level domain. This principle is necessary to eliminate disputes with possible registrants or trademark holders. Since any name has only one potential owner, that owner has the option of using it and paying any associated fees, or not. No other registrant would be allowed to use that name, so there can be no conflict between multiple potential registrants desiring the same domain name. This requirement also eases the coordination problems between multiple registries sharing a top-level domain. Since any name can only be claimed by one registrant, there is no danger of different registries promising the same name to different registrants. This requirement may seem impossibly restrictive, but a little creative thinking quickly turns up many ideas for naming schemes that provide unambiguous ownership. For instance, second-level domain names with unambiguous owners could be constructed from telephone numbers, or from corporate legal names combined with identification of the jurisdiction of incorporation to ensure uniqueness. It is also important to note that each domain having only one potential owner does not mean that each owner has only one potential domain. For instance, a telephone number-based registry could register under the registrant's choice of the number or a mnemonic made from the number using the letters on a telephone keypad in the usual way. This permits considerable scope for marketing invention by the registrant while still having an unambiguous owner for every domain. Registries could register second-level domains only to themselves and then register clients at the third level - without a requirement for unambiguous ownership. This could become difficult to coordinate in a shared registry environment; one possible solution would be to give each registry sharing the iTLD a separate second-level domain under an unambiguous system, and then they could each register clients at will under their second-level domain. A solution for the unambiguous ownership issue would be proposed in the proposal for each new iTLD; it need not be resolved once and for all nor in the same way for every new iTLD. 5.1.1. Example of a manageable top-level domain: .rtv Consider a hypothetical international top-level domain, .rtv, with the following charter: "The .rtv domain exists to register domains associated with licensed radio or television stations, according to their callsigns. Every licensed radio station in the world has a globally unique case-insensitive alphanumeric callsign assigned by national government under international agreement. The legal holder of a callsign is eligible to register a second-level domain under .rtv with the callsign as the name of the second-level domain, as 'w1aw.rtv' for the American Radio Relay League's amateur station, W1AW. Such domains must be relinquished when the callsign is transferred or no longer licensed." This domain is manageable because it uniquely specifies one possible user for any given second-level domain. Prospective registrants who wished to debate the right to a particular domain would first have to go through the appropriate legal processes to dispute ownership of a radio callsign. The registry for .rtv would not be required to resolve such disputes unless it happened to be a legal body responsible for callsigns. This international top-level domain has the undesireable characteristic of being flat, with potentially many thousands of second-level domains. However, that problem would be faced by the delegate registries that chose to operate the domain, and is not a burden on the global root servers or similar systems. 5.1.2. Example of an unmanageable top-level domain: .inc Consider a hypothetical international top-level domain, .inc, with the following charter: "The .inc domain exists to register domains associated with corporations, as an alternative to geographic domains which have names less marketable to the general public. Legally incorporated bodies with names ending in 'Inc.' or 'Incorporated' are eligible to register second-level domains under .inc based on their names, as 'acme.inc' for a corporation named 'Acme, Incorporated'." This domain is unmanageable because it does not specify a unique potential registrant for every second-level domain. Corporate names are usually unique within a country or other jurisdiction, but they are not globally unique and so disputes could arise when more than one "Acme, Incorporated" wants the "acme.inc" domain. 5.2. Shared registration All new international top-level domains must be shared between all interested and capable registries. Any organization proposing a new international top-level domain must submit a proposal showing how the new domain fills a need not served by currently existing domains. The proposal must include a charter similar to those exemplified above, though likely much longer and more specific, defining the one entity eligible to register for each second-level domain. Since top-level domains are a globally unique and limited resource, there is likely to be desire for more of them than the global network's cacheing facilities can support. No organization but a national government is big and important enough to deserve exclusive ownership of a new top-level domain. Even if individual servers can support millions of domains, the DNS architecture cannot. For this reason, any proposal for a new international top-level domain must include provision for sharing the responsibility of the domain with all capable and interested registries. Proposals must include detailed technical descriptions of the protocols and facilities to be used to make the database available to potential registries. These descriptions will be evaluated along with the other material in the proposal. 6. Summary of options available to new registrants and registries To summarize, here are the options available to an organization such as a business which wants to register a new domain name. From the registrant's point of view, the name should be memorable and usable as a trademark. From the global network's point of view, the name must be unique, descriptive, and equitably assigned. These options are in order of decreasing preference. 6.1. Geographic registration The preferable form of registration for all organizations is under a two-letter "ISO" domain name administered by a national government or delegate. Geographic registration neatly moves the demands of dispute resolution and administration away from the roots and to local agencies which may use policies and procedures appropriate to their nation's culture and practices. The disadvantage is that for globally-active organizations, it may be undesireable to be linked to only one country. Too, national governments may institute rules for local registration unfavorable to some registrants. 6.2. Third-level iTLD registration under a delegate registry Any organization may register a third- or lower-level domain name under a delegate registry in one of the original international top-level domains. Such registration is subject to the policies and procedures of the delegate registry. Third-level registration allows use of current international top-level domains in a manageable fashion. No one registry must handle the many thousands of potential registrants under a domain like .com. Instead, the .com registry handles the second-level registries, which very likely would not number more than a few hundred, and each second-level registry would handle as many customers as it can. Second-level registries would be free to choose marketable or topical names, which would lead to more descriptive information in a domain name than currently exists. Although second-level registries under current iTLDs would be granted exclusive ownership of a name, this form of monopoly would be less galling to registrants than assigning exclusive use of top-level domains, because of the much larger number of available second-level registries and the ever-present option of geographic registration. When third-level registration under iTLDs or geographic domains becomes the norm, the alleged stigma on "long" and "unmemorable" multilevel domains will be alleviated. 6.3. Second-level registration under a shared registry Second-level domains under new international top-level domains are available to any organizations meeting the requirements set out in the charter of the top-level domain. Ordinarily this would include a limit on the number of second-level domains available to each organization. Organizations could freely choose which registries to do business with. Author's addresses Matthew Skala Ansuz BBS 5760 Sooke Road Sooke, BC, CANADA V0S 1N0 voice: (250) 642-3607 data: (250) 642-7820 email: mskala@ansuz.sooke.bc.ca draft-skala-dnsreform-00.txt Expires May 30, 1997 M. Skala