|
|
|
|
|
|
|
The problems of long, complex Internet addresses and unwieldy search-engine results are being eroded by a new search method. Stuart Mark explores the latest advances in name resolution technology. Common Names Resolution Protocol The Internet has established itself as part of everyday life. It is faster, can be accessed cheaply and provides a much improved quality of site and service. But amid all this progress, the means by which a web site is located and contacted is still extremely cumbersome. There is nothing more frustrating than typing a long website address full of symbols and meaningless character sequences into the ‘Open’ or ‘Address’ field of your browser. The alternative of using a search engine isn’t much better as, with all but the most stringently constructed search strings, the result can be thousands of sites, many of which are duplicated or bewilderingly obscure. Finding A Site Each website on the Internet is identified by a Universal Resource Locator (URL) or Univesal Resource Identifier (URI). This contains a domain name which has a familiar format to most people, especially since the advent of the dot com business model. The names are held on Domain Name System (DNS) servers scattered around the web and their hierarchical dot-notation format allow them to have geographic and functional significance as in .uk and .org. Each name must be unique and this is the first problem with URIs. Many organisations and individuals find that their first choice name has already been used and have to settle for something less fitting and, in many cases, a name that doesn’t effectively describe the service they offer. The second problem with the URI is that it can easily turn into a typing nightmare. In its most basic form, a URI will contain just a DNS name and will provide access to a site’s homepage, for example www.itweek.co.uk. If, however, direct access to a specific page deep within the site is required, the URI will contain the DNS name followed by the file pathname or script. If the URI of a site is not known, a search engine must be used. These are ubiquitous on the web and allow a search to be carried out on any word or phrase. The need for such a facility is obvious but the way in which most search engines work severely restrict their effectiveness. In order for a website to be located by a search engine, the site owner must have first registered the site with the search engine operatator. In most cases, reqistration involves the notification of the URI after which a ‘Crawler’ or ‘Spider’ robot program will interrogate the site. The information gathered will be used by subsequent searches. This means that a search based on a specific word will locate all sites containing that word or, even worse, all pages within the same site containing the same word. This is why a search will often return numerous instances of the same URI. Common Names To address the problems of website search and access, the concept of common names resolution has been formed and is gaining industry-wide acceptance. A common name is an easily recalled identifier that can be used to provide direct access to a website. The idea is that a site can be located simply by typing a keyword or phrase into the ‘Open’ or ‘Address’ field of a browser instead of a URI. Alternatively, a common names website can be used to perform a search. Common names are held in a database and reference specific areas of a particular website. The difference between a search-engine search and a common name search is that the common name method does not actually search websites but instead relies on the database to return results. This allows much closer control of search parameters and more accurate results. A number of commercial common names solutions are already available. Netword invites subscriptions from web organisations for a yearly fee. A company can choose brand, product, service or any other type of name and reference it to any location in their website. Prospective customers can search from the Netword website or download a free plug-in for their browser to allow searches from the address field. Even more impressive is the offering from RealNames. Again companies register names for a fee but anyone with Internet explorer 4.x or higher or the NeoPlanet browser can perform common names searches direct from their browser without the need for a plug-in. This is because RealNames has formed pertnerships with Microsoft for browser functionality and with a numerous web portals and search engines including MSN, Google and Alta Vista. If you use IE, try entering a word into your browser; Microsoft, Novell, Nasa and even Pokemon will connect you directly to these websites. Even spaces and symbols can be used. Mutiple hits will take you to the MSN search site and RealNames hits will be highlighted with a special icon on the results page. Both Netword and RealNames use proprietary names databases and transport methods but, such is the potential for this technology in the context of ecommerce, B2B solutions and a more efficient marketplace, it has become the subject of an Internet Engineering Task Force (IETF) working group. CNRP The Common Names Resolution Protocol is intended to become a standardised method of request-repsonse connections using common names. It will use what the (IETF) refer to as Parameterised Resolution which means that flexible descriptors will be used as query parameters to resolve specific URIs. These descriptors will be permitted to contain the symbols and different language characters available from UTF-8 encoding instead of the 37 character ASCII limitation imposed by DNS. XML will be used as a transport on TCP port 1096. Also the shorter names should benefit wireless applications. Informational RFC 2972 lays out the design goals for CNRP and the framework for a Global Registry has recently been announced by the Internet Corporation for Assigned Names and Numbers (ICANN), part of which should provide a standard repository for Common Names. This could replace the proprietary databases used at present using CNRP for communication with search clients. Logical centralisation of the registry may be achieved by a CNRP referral model which would share name information between databases in much the same way that DNS forwarding works today. Namespaces are intended to be the method of storing names. These are defined by the RFC as sets of common names within a category such as Movie Titles under Entertainment or Company Names under Plumbers under Home Services. CNRP is required to include simple search filter functions which, when used in conjunction with a set of predefined categories, will greatly reduce the size of search results compared to those from search engines. One other advantage of categorisation is name duplication. Where URIs must be unique, identical common names can be registered by different organisations. If a name search finds duplications, it can return results listed by category. At present, only one Informational and two Internet-Draft RFCs exist, but the latter are expected to move to Proposed-Standard status shortly. The most widely available implementation of Common Names Resolution, that from RealNames, is likely to be one of the first commercial implementations of CNRP. The company is working with the IETF in the development of CNRP and has recently announced a restructuring in response to the advent of the Global Registry. However, there is little evidence of interest from the major browser suppliers. Microsoft’s partnership with RealNames should force CNRP into IE but Netscape may not follow until business pressures prove too great. Either way, the days of wandering aimlessly around the web appear to be numbered. www.realnames.com www.netword.com www.ietf.org www.icann.com www.msn.com Key Points
|
This site was last updated 04/25/07
