David Manura

CSE-498 (Adv. Networks)

2003-01-30

 

Development of the Domain Name System—A Review

 

The paper “Development of the Domain Name System” (Mockapetris and Dunlap) assesses the design and characteristics of the Domain Name System (DNS) five years after the system’s initial design.  The paper looks at the design, design rationale, usage patterns, production environment experience, and lessons learned.  The paper is practical, rather than theoretical, in that political and human behavior issues are also addressed.

 

The paper is relevant.  Although DNS had stabilized such that drastic overhaul of DNS is ruled out, the ideas presented can assist Internet designers in tuning the performance and usability of DNS.  Moreover, the problem of naming is presented in a much more general and still open context (e.g. file naming and X.500), in which case the experience with DNS remains highly relevant, at least as a proven baseline.

 

However, what essential problem does DNS intend to solve?  This is neglected. Yes, DNS replaces HOSTS.TXT, but why is name resolution needed?  What is the implication of an Internet that has no DNS or HOSTS.TXT but instead performs name resolution at a higher network level or a more granular level?  What should be the scope of name resolution?  Indeed, DNS operates logically with respect to the IP layer: the naming service maps one or more hierarchical names (independent of network topology) to a single IP address.  Yet one could instead map host names to multiple IP addresses, which is used in round-robin DNS to achieve scalability (Brewer, 2001).  Conversely, when host names for different web sites resolve to the same IP, the final resolution must occur at the application level (as is done in HTTP to support virtual servers with a single IP address).  Names might alternately be resolved with respect to UDP/TCP or higher network level, in which case a name could resolve to a host and port (i.e. an OS process).  This would allow URLs such as http://www.myserver.com:8080 to be replaced by a friendlier http://www2.myserver.com.  More finely granular schemes are also possible (e.g. objects in a distributed component architecture).  Does DNS serve to provide human friendly names?   Such names have minimal utility: people can use search engines and locally managed name mappings such as bookmarks.  With the commercialization and trademark battles on the Internet, some have found the “human-friendly” names still inadequate, and hence non-ASCII domain names and the failed RealNames (Winer, 2002) were born.  I believe that the main utility of DNS is to provide a level of indirection and hence name permanence transcending physical network.  Yet indirection is not mentioned in the paper.  However, even when using domain names, resource URLs often become broken, not only because resources are deleted but because URLs and similar identifiers (e.g. file extensions) change.  This could be a limitation of a ubiquitous partitioned naming system.  Many of these high-level topics are missing, largely due to the paper’s age.

 

The authors speak from experience and recognize non-technical concerns, such as user acceptance and human behavior, concerns poorly addressed in many other papers.  Positive contributions include the discussions about uncontrolled changes to externally developed software, “loss of interest” by developers in maintaining basically working software, administration, and user resistance.

 

Finally, the intended usage of and reason for type and class entries is unclear.

 

In summary, the paper is informative, thought provoking, and clear but at times dated and missing important issues.

 

 

---

Brewer, E.  “Lessons from Giant-Scale Services. IEEE Internet Computing.”  July-August 2001.  pp. 46-55.  DRAFT: http://citeseer.nj.nec.com/476298.html

 

Kubiatowicz, J. et al.  “OceanStore: An Architecture for Global-Scale Persistent Storage.”  In Proceedings of the Ninth International Conference on Architectural Support for Programming Languages (ASPLOS 2000), November 2000.  http://oceanstore.cs.berkeley.edu/publications/papers/pdf/asplos00.pdf

 

Mockapetris, P. and Dunlap, K.  “Development of the Domain Name System.”  In the Proceedings of SIGCOMM ’88, Computer Communications Review, 18, 4, August 1988, pp. 123-133.  http://www.acm.org/sigcomm/ccr/archive/1995/jan95/ccr-9501-mockapet.html

 

Winer, D.  “RealNames and Microsoft.”  ScriptingNews.  http://scriptingnews.userland.com/realNamesMicrosoft.  11 May 2002. (from the Google cache)