From: Susumu Harada (harada@cs.washington.edu)
Date: Wed Nov 17 2004 - 06:54:48 PST
"Development of the Domain Name System"
P. Mockapetris and K. Dunlap
This paper provides a historical context of the initial motivations behind
the design of the Domain Naming System. As the HOSTS.TXT mechanism grew
inadequate to meet the growing need for a more scalable solution, the DNS
design was proposed to provide a highly distributable and expandable
system . DNS stores its name space data as a trie of substrings where the
names can be recovered by concatenating the substrings along a path in
this trie. This is further subdivided among multiple zones, where each
zone is controlled by particular organizations, who are responsible for
the propagation of appropriate updates to the trie. This structure
enabled the DNS system to scale and be distributed across multiple
I thought the notion of negative caching was clever, although it would
seem that the cache size would increase significantly by including such
cache set, and without any details on how they are actually implemented,
it is difficult to see what other limitations it may have.
There were several open issues that were not addressed by the authors.
For example, they stated that "at present, we estimate that 50% of all
root server traffic could be eliminated by improvements in various
resolver implementations to use less aggressive retransmission and better
caching," however do not propose any concrete suggestions on how to go
about achieving their proposed effect.
There also seems to be many robustness issues as well. For example, they
state that the root domain is supported by seven redundant name servers.
Although it is a level of redundancy, there is no indication of how they
came up with that number to what extent they expected it to scale. Also,
their delegation of zones to various organizations seem to give too much
power to uncontrollable entities, leading to a possibility of degraded
level of service if those organizations do not conform to certain
performance guidelines.
An area that I would like to see investigated more is in domain name
resolution in the context of mobile hosts, and the analysis of how the
current solution is able or unable to scale to such dynamic environment as
more and more mobile devices proliferate.
This archive was generated by hypermail 2.1.6 : Wed Nov 17 2004 - 06:54:48 PST