From: Scott Schremmer (scotths@cs.washington.edu)
Date: Wed Oct 06 2004 - 08:02:29 PDT
This paper outlines the Internet protocol as it was being created
in the mid-1970's. At the time, this protocol was new and untested. The
paper gives a good overview of the protocol and how it deals with issues
such as addressing of packets, reliability through retransmission, dealing
with different types of networks, etc. The paper shows an impressive
amount of foresight given the lastingness of this protocol.
On the negative side, the paper presents only the proposed protocol
without actual results or simulation. Of course the test of time as shown
that this protocol works quite successfully. This paper doesn't take into
account the need for unreliable communications which would later lead to
the split of UDP and IP from TCP. Lastly in a comment that amusingly
echos the famous Bill Gates comment regarding computer memory the authors
state that 256 networks of 65,536 ip addresses should be adequate.
As many of the details of the proposed protocol made it into the currently
used TCP/IP protocols, having an understanding of the historical reasons
for the design decisions could be helpful in making current decisions
regarding use and modification of the protocol.
This archive was generated by hypermail 2.1.6 : Wed Oct 06 2004 - 08:02:29 PDT