Keywords Naming Services BOF (kwns) Thursday, March 21 at 1300-1500 =================================== CHAIRS: Yves Arrouye Yang Woo Ko Abstract This BOF will examine so-called Keywords systems. Keywords systems provide multilingual, human-friendly, natural language naming systems for Internet applications. This BOF will use a collection of documents relevant to the topic, some of which were produced by Keywords systems operators, or relevant to the topic of Internet navigation, as the starting point for any subsequent working group or other efforts. Introduction There are several diffuse efforts working toward the goal of offering an internationalized multilingual, human-friendly, naming system for Internet applications and their users. The two common answers to the problem are directory services [2, 3] and Keywords systems [4, 14]. Keywords are a layer above DNS and search systems, but they are not a directory layer or service (see [3]). The uniqueness requirements of Keywords make them an unambiguous lookup system. Keywords systems are a focused first step to understand and solve some of the issues proposed in [2,3] and presented at the irnss BOF at the 52th IETF meeting. There are many Keywords systems in existence, or about to be brought into existence, today. Companies like RealNames Corporation [5], Netpia [6], Netword [7], AOL [8], CNNIC [9], TWNIC [10], and 3721 [11] are operators of Keywords systems. All these systems try to solve the same problem. They are turning to the IETF for their standardization issues. There is a need to satisfy a common set of requirements for Keywords systems, in order to enable interoperability between the different Keywords systems operators through standardization of the resolution and registration protocols. In order to understand the difference between Keywords systems and the proposed directory and search solution, we propose to discuss the following items as initial requirements that stem from the existing approaches to Keywords systems: - A Keyword as a named handle to a resource descriptor that provides meta-data about a given resource, including its physical address (resource URI). At stake is the notion that a set of properties, or facets, makes up a Keyword; whether to standardize these properties for interoperability; and whether it is desirable or not to provide mechanisms to extend them. - Keywords can serve the needs of multiple applications and devices. A Keywords system may support more than one application, device, or service. Different applications, devices, or services, may require different properties in the resource descriptor. - Keywords are unique within the context defined by a well-defined subset of properties of the resource descriptors. Given a multilingual name and some context, an application can fetch a resource descriptor and use its properties, for example to provide direct navigation. - Keyword namespaces should be organized to provide for local customization and culturally correct handling of names for a given region. The issues to be discussed ranged from the geopolitical implications of the choice of the namespace granularity (e.g. country-based namespaces), to the amount of latitude that a given namespace can have in the (re)definition of name equivalence within that namespace, including, for example, handling the relationship between the Simplified and Traditional forms of Chinese. - When two Keywords systems operators want to be integrated, they need to standardize on a resolution protocol, which should be standardized. In addition, there may or may not be a need for a standardized registration mechanism for Keywords across different Keywords registry; such standardization may be based on the PROVREG WG efforts. - Applications will need to be changed in order to benefit from Keywords or any new naming system. Because changing applications is a disruptive operation, there is a strong incentive to design a new protocol that will enable a resolution system to resolve names and/or addresses from multiple naming systems through a single common application interface. - Keywords systems are not directory services or directory layers, but are an independent layer above DNS. There is a need to describe the relationship and differentiation between this standard effort and the proposed Internet Resource Name Search Service (irnss) effort, and the CNRP [13] effort, should be discussed. These requirements are presented in [14], a document that reflects the view of the members of the MINC and IFIK associations, [5, 7, 9, 10, 11, 12]. Proposed Timeline If an IETF Working Group is created after the BOF is held, we propose the following timeline for the WG deliveries: - A requirements document will be produced within 6 months. - A resolution protocol for Keywords will be defined within 12 months. - Timelines for other items will be proposed as a result of the BOF, as these items are determined to be potential WG work items. BOF Agenda 1. Existing Keywords systems overview. Each of the following systems will be presented in light of the proposed discussion items. Each presentation will last 5 minutes. 2. Discussion of the items presented above, including emphasis on multilingual issues such as TC/SC, which hasn't been successfully resolved so far. 3. Decision on next steps: working group creation, and if so, timelines (i.e. when to produce different documents). References [1] draft-klensin-dns-role-01.txt [2] draft-klensin-dns-search-03.txt [3] draft-mealling-sls-00.txt [4] draft-arrouye-kls-00.txt [5] http://www.realnames.com/ [6] http://www.netword.com/ [7] http://www.netpia.com/ [8] http://www.aol.com/ [9] http://www.cnnic.net.cn/ [10] http://twnic.net.tw/ [11] http://www.3721.com/ [12] http://www.ifik.org/ [13] Popp, N., M. Mealling, L. Masinter, K. Sollins. "Context and Goals for Common Name Resolution", RFC 2972. October 2000. [14] Arrouye, Y., and Tan, T.W.. "Keywords Systems - Definition and Requirements", draft-arrouye-keywords-reqs-00.txt, February 2002.