From: Tom Christiansen (tomchr@ee.washington.edu)
Date: Wed Nov 17 2004 - 02:18:55 PST
This is a historical perspective paper which describes the motivation
behind the development of the DNS. In The Good Old Days, a list of domain
names were maintained in form of a HOSTS.TXT file. This file was present on
all hosts in the internet and was managed from a central point (at SRI
Network Information Center). Needless to say, as the number of hosts on the
Internet grew, so did the need for a different system. The biggest change
was to make the system distributed. In addition, the main requirements
involved that the system should be OS independent and have at least the
same functionality as the old HOSTS.TXT system. The DNS should also have no
obvious limits for the size of names. Four years after the development of
DNS, the DNS name list contained about four times the number of entries as
the HOSTS.TXT did. This is some measure of success.
The paper mentions some issues regarding look-ups of names that don't
exist. A majority of these look-ups are caused by users on other
internetworks trying to see if a particular name is accessible through the
Internet. If the name formats for the different internetworks was different
it would have been simple to set up a filter to eliminate the unwanted
queries. Maybe this wasn't the case, who knows...
It is interesting to note that the fundamental structure of DNS naming has
not changed. It's still a hierarchical structure. The TLD naming (.com,
.edu, .mil, etc) has not changed either, although, new ones have been added
as needed.
Another interesting point is that many of the programs, protocols (MX,
BIND, etc) mentioned in this article are still in use. Linux even supports
the HOSTS.TXT system - highly usable on small local networks.
The notion that documentation should be written with the assumption that
only the examples will be read is dead on.
This archive was generated by hypermail 2.1.6 : Wed Nov 17 2004 - 19:53:20 PST