Date: Tue, 13 Jul 1993 17:35:46 +0100 (BST)
From: Martin Hamilton <M.T.Hamilton@lut.ac.uk>
Subject: URNs in the DNS
To: uri@bunyip.com, iiir@merit.edu
Message-Id: <Pine.3.07.9307131746.D913-d100000@lust>
..so where's Eric? :-)
This is about as far from an ideal solution to the URN directory
problem as you could get, but I think it's a good pragmatic response
in the Internet tradition. Comments and/or flames welcome!
Cheers,
Martin
--
URNs in the DNS
1 The Whyfores
This document describes an interim approach to the provision of the URN to URL directory service described in [9], and goes on to show how this may be used to provide a partial solution to the problem of file caching in a global Internet.
It is the author's contention that the Internet Domain Name System (DNS) may be used as is in the role of URN directory, by making use of the TXT Resource Record (RR) encoding of arbitrary attributes described in [6]. Should this experiment prove successful, it is envisaged that a formal RR for the URN to URL mapping might be defined.
Sample source code is presented as appendices for a) a simple URN lookup program and b) a program to generate the appropriate TXT records automatically from the results of an archie [3] search.
2 Jargon
URNs - from [8]
A Uniform Resource Name (URN) is an identifier which can be used to uniquely identify a resource and is designed to provide persistent naming for networked objects. This name would stay the same no matter what the current location(s) of the object was.
URLs - from [2]
Current names are basically location specifiers (addresses). These may be known as Uniform Resource Locators (URLs). They give the necessary parts of an address for a reader to access an information provider using the given protocol, and ask for the object required.
DNS - from [5]
The domain name system is a mixture of functions and data types which are an official protocol and functions and data types which are still experimental. Since the domain system is intentionally extensible, new data types and experimental behaviour should always be expected in parts of the system beyond the official protocol.
3 domain authority in URNs
[8] describes an encoding for URNs which looks like this:
URN:Naming_Authority_identifier::opaque_string::: ^ | ^ | ^ |_________________ wrapper ___|_______________|
This paper proposes that the authority registration domain be used as shorthand to imply that the object referred to exists as an entry in the DNS, and may have associated TXT resource records containing the URLs of network accessible copies of it, encoded in the form described in [6].
4 Example
So, if <URN:domain::finger.foobar.lut.ac.uk:::> were to be used as a identifier for the latest version of the GNU Finger program, the placement of the following entries in the DNS for the domain finger.foobar.lut.ac.uk
finger TXT "URL=ftp://sunsite.unc.edu/pub/gnu/.cap/finger-1.37.tar.gz" finger TXT "URL=ftp://theta.iis.u-tokyo.ac.jp/pub1/GNU/finger-1.37.tar.gz" finger TXT "URL=ftp://utsun.s.u-tokyo.ac.jp/ftpsync/prep/finger-1.37.tar.gz"
would cause a URN lookup to result in:
URL=ftp://sunsite.unc.edu/pub/gnu/.cap/finger-1.37.tar.gz URL=ftp://theta.iis.u-tokyo.ac.jp/pub1/GNU/finger-1.37.tar.gz URL=ftp://utsun.s.u-tokyo.ac.jp/ftpsync/prep/finger-1.37.tar.gz
The DNS client is then free to sort these results according to some predetermined algorithm, e.g. by geographical weighting.
5 Conclusions
There is a pressing need amongst the Internet community for a widely available, prefably decentralised, means of managing URNs and associating them with URLs. This is necessary to avoid the problems associated with temporary names (URLs), which are stored in directory services such as archie and would be helpful in providing support for rudimentary file caching on the Internet, as it provides a mechanism which is suited to the automatic selection of geographically ``close'' copies of a widely distributed resource.
Note that no attempt is made to define any metadata attributes for these objects beyond those which may be derived or implied from the URN and URLs.
Likewise, no organisational scheme is described, though some variant of the aliasing scheme for multinational organisations in the OSI Directory outlined in [1] may be appropriate.
Paradoxically, URN repositories may be bootstrapped from automatic searches of existing directories, e.g. archie - the information provider need only do this periodically, making the results available to everyone.
Finally, it remains to be seen whether the DNS infrastructure is capable of coping with these new demands. In particular, at least one major vendor's distribution of the nameserver software does not support the TXT RRs, and it remains to be seen what limitations are commonly placed on their sizes. Likewise there is the problem of Internet address exhaustion, though this may already have been solved by the NAT effort [4] and related activities.
References
[1] Paul Barker and Steve Hardcastle-Kille. Naming Guidelines for Directory Pilots. Internet RFC 1384, University College London and ISODE Consortium, January 1993
[2] Tim Berners-Lee. Uniform Resource Locators. *** Internet Draft ***, World-Wide Web Project, CERN, March 1993.
[3] Alan Emtage and Peter Deutsch. archie - An Electronic Directory Service for the Internet. In Winter Usenix Conference 1992, pages 93--110. Usenix Association, Berkeley CA, 1992.
[4] Van Jacobson. LNAT - Large scale IP via Network Address Translation. *** Draft ***, Lawrence Berkeley Laboratory, January 1992.
[5] Paul Mockapetris. Domain names - concepts and facilities. Internet RFC 1034, Information Sciences Institute, November 1987.
[6] Rich Rosenbaum. Using the Domain Name System to Store Arbitrary String Attributes. *** Internet Draft ***, Digital Equipment Corporation, April 1993.
[7] Larry Wall and Randal L. Schwartz. Programming Perl. O'Reilly & Associates, Inc., 1990.
[8] Chris Weider and Peter Deutsch. Uniform Resource Names. *** Internet Draft ***, Merit Network, Inc. and Bunyip Information Systems, Inc., May 1993.
[9] Chris Weider and Peter Deutsch. A Vision of an Integrated Internet Information Service. *** Internet Draft ***, Merit Network, Inc. and Bunyip Information Systems, Inc., March 1993.
Appendices
A Generation of TXT RRs from archie searches
This Perl [7] script constructs TXT RRs in a particular domain derived from the results of an archie search for the specified object, and sorted to contain the ten most recent matches. It makes use of the eponymous command line client - archie.
If it were saved as urnpoke, producing the data used in the above example would be a matter of typing something like:
urnpoke finger.foobar.lut.ac.uk finger-1.37.tar.gz
The actual script is as follows:
#!/usr/local/bin/perl $archie="/usr/local/bin/archie"; $host="archie.sura.net"; $target=$ARGV[1]; $domain=$ARGV[0];
chop(@archie=`$archie -ltm10 -h $host $target`); foreach(@archie){ ($date,$size,$host,$dir) = split; print "$domain\t\tTXT\tftp://$host$dir\n"; }
B URN to URL lookup
This Perl script exploits the popular Domain Internet Groper tool - dig - to carry out the actual DNS lookup, and then filters out the TXT records which do not point to URLs.
If it were saved as urnpeek, producing the results used in the above example would be a matter of typing something like:
urnpeek finger.foobar.lut.ac.uk
The actual script is as follows:
#!/usr/local/bin/perl $dig="/usr/local/bin/dig"; $domain=$ARGV[0];
chop(@pit=`$dig txt $domain`); foreach(@pit){ next unless /$domain\.\s+\d+\s+TXT\s+(.*)/; print "$1\n" if $1 =~ /^URL=/; }