3-2 Implementation of Japanese Domain Names

(IDN-TF)

What is a Japanese Domain Name

Japanese domain names are a component of the general-use JP domain name (please refer to Special Issue 3-1 of this newsletter), in which kanji and kana characters can be used for the second level of a JP domain name, in addition to traditional alphabets, numbers and hyphen. In other words, within the general-use JP domain names, Japanese domain names can include Japanese characters such as kanji and kana in the second level domain. Here is an example of a Japanese domain name.

A Japanese domain name is created through multilingual domain name technology, which makes it possible for different types of letters to be used in addition to traditional domain names. It is not the case that JPNIC, on its own, created a hierarchical structure or a name space to the existing domain name.

Trend Toward Standardization of Multilingual Domain Names

Multilingual domain name technology is going through a standardization process by the IDN-WG (Internationalized Domain Name Working Group) of the IETF (Internet Engineering Task Force.) The multilingual domain name system is one of the domain name internationalization technologies, and adds characters that can be used for in traditional domain names.

IDN-WG is a working group, newly established in February 2000, that is working hard on standardization. However, it has not yet progressed to the point that an RFC (Request for Comments, documents which state the Internet standards) can be published.

The work of the IDN-WG is based on the following principles:

• Multilingual domain names must not influence current DNS use and management.
• Multilingual domain names should continue to be DNS that can resolve names in any type of system, regardless of the location.

Work in progress includes the following:

• Summarization of international domain name requirements.
  -> There are plans to publish the results as an Informational RFC.
• Creation of a format that represents multilingual domain names in the DNS (Domain Name server) protocol.
  -> There are plans to publish the results as a Standard track RFC.

Format of the multilingual domain name on the DNS protocol is under discussion based on the following policies.

• With regards to character sets, the UCS (Universal Multiple-Octet Coded Character Set) of Unicode (ISO/IEC 10646) is used.
• In order to maintain compatibility with existing DNS, ASCII Compatible Encoding (ACE, hereafter) will be used. This encoding will represent multilingual domain names only in ASCII characters, which are currently used in domain names.
• Normalization of multilingual domain names (such as unification of double byte and single byte, and of upper case and lower case characters) shall be conducted before converting to ACE.

With these points in mind, normalization rules are also being discussed. Regarding the representation format and normalization rules, a special team was organized within the IDN-WG to review the currently proposed formats in detail. We will report on the trends in multilingual domain name standardization on a regular basis.

JPNIC Policies

JPNIC's policies regarding multilingual domain names have not changed since "JPNIC's Approach Regarding Multilingual Domain Name (draft)"

http://www.nic.ad.jp/jp/topics/archive/20000225-01.html

was published on the JPNIC web page on February 25, 2000. The following summarizes the conditions for introduction as stated in the policies.

1. Technical and Managerial Conditions
   1. ZLD^(*), otherwise alternative multilingual domain name space schema, should be decided and well coordinated within the whole Internet.
   2. DNS protocol, especially the character code system, should be decided.
   3. There should be agreement within gTLD and ccTLD, and rules, regarding multilingualization of the top-level domain name.
2. Implementation Conditions
   1. Implementation of a standard DNS server (such as BIND) which meets technical requirements.
3. Operational conditions
   1. Care to non-Japanese users. It is necessary to pay special attention to non-Japanese language users. Such care includes setting aside an alphabetic domain name that corresponds to the Japanese domain name.
   2. Many disputes over trademarks may occur, thus methods to resolve these disputes must be prepared.

More than six months have passed since these policies were published. The conditions have been met as shown below.

1. 1. By proposing to use ACE, which has an identifier, it became possible to represent multilingual domain names within the current domain name space.
   2. There is almost a consensus regarding adoption of UCS.
   3. Because multilingual domain names are now represented within the current domain name space, the issue of multilingualization of TLD has become a long-term issue.
2. 1. The use of ACE made it possible to operate with the existing DNS.
3. 1. The use of ACE solved the problem of terminals which can not support multilingual domain names.
   2. "JP Domain Name Dispute Resolution Policy" was enforced.

http://www.nic.ad.jp/jp/regist/dom/drp/

With the above developments, JPNIC decided to start test operation of Japanese domain names, based on the policies stated in the "Technical Policies regarding Japanese Domain Name within General-Use JP Domain Name System," which was published on the JPNIC web page on October 10, 2000.

http://www.nic.ad.jp/dotjp/doc/tech-policy.html

Here is a summary of conditions for the technologies to be adopted, as stated in the Technical Policies.

• We will follow IETF IDN-WG policies.
• We will not adopt any format which requires ZLD, as the format to realize multilingual domain names. This is due to the fact that the ZLD method may divide the domain name space.
• In order to minimize the influences on the current DNS and applications, we will adopt ACE for character encoding on DNS protocol. However, we will not adopt ACEs that cannot be distinguished from regular ASCII characters, and those that require ZLD.
• RACE (Row-based ACE) will be used as the ACE during the test operation.
• Resource records for setting Japanese domain name on the JP zone name server shall be NS.

"Technical Rules of General-Use JP Domain Name Registration," published on November 2, 2000, describes Japanese domain name registration in detail, as to available characters, the length of the character string and normalization rules (process to avoid the same characters literally or semantically from being mistaken as different characters).

http://www.nic.ad.jp/dotjp/doc/saisoku-1.html

Here is a summary of the detailed rules.

Available characters

First level Kanji characters, second level Kanji characters, Hiragana, Katakana, and the seven symbols indicated below.

In addition to the above, alphabet, numbers, hyphen are available. (Hyphens can only be used in the middle of a string, and not as the first or last character).

Character string length.

More than 1 character and not more than 15 characters
(When at least one Kanji character, Kana, or symbols are used).
More than 3 and not more than 63 characters
(When all characters consist of only alphabets, numbers, and/or hyphens.)

Normalization rule

ASCII is used for alphabetical and numerical characters.
Double byte characters are used for Kana.

Representation Format for Japanese Domain Names

As mentioned above, RACE will be used for Japanese domain names during the test operation.

RACE is based on the UTF-16, which is one of the UCS encoding methods. Its features are as stated below.

• It has ACE identifiers.
  (RACE version 3 uses a prefix of BQ--. There is no distinction between upper case and lower case characters for BQ)
• It has a row-based compression algorithm.
  (When character code is within a certain range, this format iterates a common bit pattern).

Examples for representation of Japanese domain names by RACE are shown below.

JPNIC Operation Test

JPNIC has planned a two-phase operational test of the Japanese domain name according to the previously mentioned "Technical Policy regarding Japanese Domain Names within the General-Use JP Domain Name System," and started Phase 1 on November 6. For further detail of the Phase 1, please refer to the JPNIC web page.

http://www.nic.ad.jp/jp/research/idn/index.html#phase1

Here is a summary of each phase.

Phase 1

Japanese domain names which JPNIC reserved for the test are set on the JP zone name server and an interoperability assurance test is being conducted. This phase provides a test environment for the pre-evaluation of the access to the multilingual domain name system.

Phase 2

Japanese domain names registered by users as general-use JP domain names will be set on the JP zone name server. Comprehensive operational tests will be conducted. This phase will see the compatibility of various applications that refer to the multilingual domain name system and will promote acceptance of multilingual domain name.

It is scheduled to start April 2, 2001, but this is subject to change. Phase 2 is scheduled to continue until the IETF IDN-WG completes the standardization of multilingual domain name character encoding (Issuance of Proposed Standard RFC.)

What does it Mean to be Able to Use Japanese Domain Names

JPNIC's Multilingual Domain Name Evaluation kit (mDNkit, hereafter) provides a test environment for resolving problems regarding two-way compatibility and conversions of names and addresses on multilingual domain name DNS. It does not necessarily ensure an environment in which all applications can support Japanese domain names. Here is an overview of what it means to be able to use a Japanese domain name.

With this in mind, let's first look at where domain names are being used. Generally speaking, domain names are used in mail addresses (such as foo@example.co.jp) and in URLs of WWW (http://www.example.jp/). It is also used in a situation where you designate the computer of your counterpart. (For example, telnet host.example.co.jp, ftp ftp.example.co.jp).

Thus the usage below can naturally be considered.

However, the use of domain names goes beyond the above. For example, many current UNIX systems use domain names as indicated below. UNIX systems

% hostname
host.example.co.jp

These are not related to DNS.

It is ideal if Japanese can be used in all of these situations. But realistically, it may not become a reality. In particular, a hostname depends on the OS, and it is impossible to do so unless the OS itself is internationalized.

The mDNkit does not deal with the latter examples, but with the former examples.We believe that most users assume that Japanese domain names mean the use of Japanese in the former arena.

However, the current situation is such that Japanese domain names cannot be used even in the former cases unless applications can support Japanese domain names. The mDNkit provides an ad hoc solution to name and address resolution within applications. However, users may not totally be satisfied with it. For example, the mDNkit is useless for e-mail software, in which a mail address is tightly bound to a protocol. In this case, radical measures should be taken to the entire mail system. In the future, of course, as modifications of applications progress, Japanese domain names can be used in various situations as described before, where traditional domain names are currently used.

However, users may not be content yet. For example, what shall we do with the "foo" part in a e-mail address such as

Using Japanese for this part raises a separate issue. Namely, how to use Japanese for an account on a mail server itself. Of course, using Japanese for an account may include using Japanese for user names on UNIX and Windows NT.

In addition, there is a problem of URL path names, as indicated in the following example.

Path names (as well as IDs) also depend on OS as described in the case of a hostname. Thus, the problem of using Japanese on OS as a whole remains a large problem. If it is a matter of ID, conversion rules capable of handling Japanese may be sufficient, but it may be more difficult than using Japanese as a domain name.

Furthermore, there is a problem with the delimiter (it would be ideal to change the delimiter to Japanese as well) as indicated in the following example.

We also have an issue of top-level domain as indicated below.

The ideal for a user may be the last example, but JPNIC would like to be careful in using Japanese for delimiters. This is because delimiters depend much on applications, and applications may or may not function. One may ask whether standardization of the delimiter is possible, but this is a local issue, and an international standard for delimiters is impossible.

The issue of top-level domains should be discussed in the system of the Internet as a whole; there is nothing that we can do in the current situation.

As you saw in the above discussion, there are many consequences of using Japanese everywhere for describing an address, and it is currently not at a level where a user can utilize it. We are at the stage where the use of Japanese (the use of multilingual system, to be exact) is being standardized, and users can only use Japanese in a domain name in the form indicated at the beginning.

Technical Details of the Multilingual Domain

JPNIC has a web page on "Technical Details of the Multilingual Domain." Please refer to this page for explanations and links to relevant information regarding technical information and the test environment as mentioned previously.

http://www.nic.ad.jp/jp/research/idn/

---

(*)ZLD: Zero Level Domain: A special top-level domain hidden from users, which identifies that it is a multilingual domain name.