(no subject)

From: Tom Walsh <tom_at_cs.toronto.edu>
Date: Wed, 13 Sep 2006 22:48:12 -0400

The paper describes the design of what would become the internet. It
describes the historical context in which this work and certain
requirements created by that context before going on to lay out the
primary and secondary goals of the internet's designers and then
discussing the approaches used to satisfy these goals and some of the
associated trade-offs.

Looking at this paper in a contemporary context, I'm amazed that they
designed with "real time delivery of digitized speech" in mind, given
that we still struggle with this application today. However there
seem to be a number of applications that were not foreseen (and as
such not designed for), primarily related to multicasting. TCP was
designed for bi-directional reliable delivery of data, while UDP was
designed for unreliable, uni-directional, but highly efficient
delivery of data. Could multi-directional delivery of data be
supported simply with another protocol complementing UDP and TCP or
is efficient multi-directional delivery a service that would require
a complementary protocol at the IP level?

The paper is clear and informative, but somewhat boring for those who
already have some familiarity with TCP/IP.
Received on Wed Sep 13 2006 - 22:48:23 EDT

This archive was generated by hypermail 2.2.0 : Wed Sep 13 2006 - 23:42:40 EDT