From: Erika Rice (erice@cs.washington.edu)
Date: Thu Sep 30 2004 - 13:56:59 PDT
Erika Rice's reivew of "The Design Philosophy of the DARPA Internet
Protocols" by David Clark:
David Clark's "The Design Philosophy of the DARPA Internet Protocols"
discusses the historical motivations for some of the design decisions
behind the Internet architecture. It also outlines some of the
strengths and weaknesses of those decisions that have come to light over
time.
The paper gives a clear outline of the goals and priorities of the
Internet architecture and discusses the rational behind the related
implementation decisions. The description of the rational is honest;
some design decisions were based on high level considerations (for
example, the decision to increase flexibility by making TCP and IP two
separate protocols) while others were based on practicality (storing
state information in end points was easier to engineer than storing it
in intermediate nodes).
However, the paper does not always present sufficient justification for
the decisions made; the paper assumes, almost without comment, that
packet switching was the right architecture choice. The only
justifications given were that packet switching is a more appropriate
model for some applications (although there are others which it is not
appropriate for) and that many of
the existing networks the Internet wanted to connect were packet
switched. Although packet switching has proved to be a flexible choice,
it would have been beneficial if the paper had considered the potential
shortcomings of this assumption in the same way that it considered the
shortcomings of some of the other design choices.
A knowledge of the motivations behind the development of the TCP/IP
architecture helps one understand both its strengths and its
shortcomings a little better. It illustrates the principle that not all
priorities can achieved well (or at all) in a first system and that it
is crucial to determine
which priorities are most important and which can wait until later.
This archive was generated by hypermail 2.1.6 : Thu Sep 30 2004 - 13:56:59 PDT