From: Kate Everitt (everitt@cs.washington.edu)
Date: Wed Nov 17 2004 - 13:35:21 PST
Development of the Domain Name System
Review: Katherine Everitt (resubmitted)
This paper presents the DNS (Domain Name System) which
still is used in the Internet today, making the design
decisions very relevant. It provides much improvement
in scalability over the HOSTS.txt system. DNS uses a
variable depth tree, with the name to a node being an
ordered combination of all the nodes above it on the
tree.
It seems to fit will with the organization of the
internet, using hierarchy to mirror topology and zones
to match how some organization has control over an
area of the network. Even when the topology
doesn’t quite follow this, the DNS system will
still work. The use of a variable depth hierarchy was
very flexible, allowing organizations to fit their
needs.
It’s quite good from the user perspective,
allowing users to be case-agnostic when typing (no
problem with caps lock) and helping them remember data
as a series of “chunks” by the 7
+- 2 rule, people are much better at remembering data
when it has been partitioned than as a set of letters
or numbers.
One of the key insights from this paper was the idea
of caching negative responses. As the network changes,
it is important to make sure we have updated data.
I was a bit worried when I saw that there were only 7
redundant name servers for the root and top level
domain. This seems like a weak point for a malicious
DOS attack – or you could cut power. These servers
are considered authoritive, so they could mess up some
caching or hide some subdomains.
This archive was generated by hypermail 2.1.6 : Wed Nov 17 2004 - 13:35:26 PST