[e2e] High Packet Loss and TCP

[David P. Reed]

At 09:59 AM 5/2/2003 -0700, Greg Minshall wrote:
>am i right that (10**-6) is order of magnitude 10% packet loss (+/-, 
>depending
>on packet size)?  if so, theory and practice and whatever seem to converge.
>always a nice (if somewhat suspicious) outcome.

One would hope that links with that kind of biterror rate were either 
Forward-Error Corrected or link-level ARQ'ed before they were inserted into 
the global Internet.   TCP, etc. are defined to work on "best efforts" 
networks, not "no efforts" networks.   Losing 10% of packets on the wires 
(or wireless links) is ridiculous.

Packet loss rates that are due to queue overflow are what TCP is supposed 
to deal with, and those are not characterized by BER, but by packet loss rates.