Review - Development of the Domain Name System

This paper, written in 1988, outlines how the need for Domain Name System
(DNS) came to be. It outlines the design considerations of a naming system
which is independent of network topology, flexible and scalable to replace
the current one in the DARPA Internet (1983). This strategy was based on a
centrally maintained HOSTS.txt file which is distributed to all hosts and
was not very scalable.

The paper is fairly well written with an outlook on the design
considerations of DNS, or characteristics that DNS ought to have to replace
currently deployed or other proposed naming systems. It then proposes an
evaluation of how well DNS, partially deployed since 1983, is doing with
respect to design decisions made regarding the variable depth name
hierarchy; the communication protocol used (datagram) and the optimizing
strategies like caching.

The authors outline some difficulties they faced in trying to educate users
to DNS. However, in my opinion, they failed to state how they got around
bigger difficulties, e.g. make software makers change their code to support
DNS because it would be pretty hard to incorporate such drastic changes
today.

Today DNS is THE established name service. The authors describe several
shortcomings of DNS and the bad performance they noticed, especially with
low speed links. They also emphasized on the design that made DNS suitable
for the current Internet. But, is it that DNS was lucky to have been
proposed at a time when workstations were getting cheap and corporations
were switching from time-sharing terminals to workstations? Is it that DNS
got lucky that improvement in communication networks made their system a
viable option? I doubt people would have been satisfied with a resolver that
takes between 3 to 10 seconds during prime time and 30 to 60seconds in the
worst case.

 
Received on Sun Oct 02 2005 - 16:49:05 EDT