From: Michelle Liu (liujing@u.washington.edu)
Date: Tue Nov 16 2004 - 22:22:05 PST
Review of "Development of the Domain Name System"
Jing Liu
This paper talks about the development of the domain name system. It serves a highly diverse community of hosts, users, and networks and uses a unique combination of hierarchies, caching, and datagram access.
The paper first discussed about the necessity of DNS. It starts with talking about HOST.TXT system, which is a simple text file, centrally maintained on a host at the SRI Network Information Center and distributed to all hosts in the Internet via direct and indirect file transfers. The existing problems with HOST.TXT leads to the development of DNS, for example, the costs of its distribution are high and centralized control of updating did not fit the distributed management of the Internet.
The basic design assumptions for DNS include: to provide enough information, to allow the database to be maintained in a distributed manner, to have no obvious size limits for information expression and to provide tolerable performance. Based on those assumptions, DNS was constructed. DNS separate the components to name servers and resolvers. The name space in the name servers is a hierarchical tree so that variable-length names are provided and hierarchical architecture is fit for the scalable property of the Internet. Database is distributed managed among the name servers. Resovers use caching to provide flexibility for queries. Datagrams are used as the preferred method for accessing name servers to restrict the traffic amount to the name servers.
The paper also talks about the shortcomings of DNS, such as data type and class grow drastically, converting network applications to use the DNS ( I doubt if this is still a big problem today, since DNS is commonly used today.), and the management of distribution control.
This paper talks about some interesting issues, such as it's easier to add a function than to remove a function, how the new architecture can be built based on the old architecture since in reality it's hard to discard all the old stuff considering cost. There are still a lot could be done in the future about DNS, such as caching consistency, caching updating frequency or how to set the TTL and distribution management of database, etc.
This archive was generated by hypermail 2.1.6 : Tue Nov 16 2004 - 22:22:29 PST