IP: TCP/IP and Satellites

[Dave Farber <farber () central cis upenn edu>]

From: Craig Partridge <craig () aland bbn com>
Date: Tue, 15 Oct 96 09:55:58 -0700


I've had a number of folks contact me about an LA Times article which appears
to say TCP/IP won't work over satellites.  (LA Times, Oct 3, business section
entitled "Satellite Firms Dealt Blow on Internet Plans" -- retrievable from
the LA Times web site).  I'm one of several people quoted in the article.


Apparently the article is causing some folks to rethink plans for delivering
Internet connectivity via satellite, which is silly.


TCP/IP usually works fine over satellites (which you might expect, given
that TCP/IP was designed with satellite networks in mind).  The one major
problem area is when satellite bandwidths get to be very high.  Then TCP
has some difficulty using the bandwidth fully.  (TCP has similar problems
on high-bandwidth terrestrial fiber).  Several researchers (many funded by
NASA) are working on this problem and all feel this is a solvable problem.
Indeed, the current challenge is not finding a solution -- a few are
known -- but rather, finding a *good* solution that also works well on
terrestrial links.


Much to the chagrin of everyone I've talked to who was interviewed for the
LA Times article, the article focusses on the problem rather than the
confidence about the solution.  That's the reporter's perogative.  But the
article's ripple effects are a reminder (a) to those of us who gave interviews
and (b) to people trying to make business decisions, that what appears in
print (or on-line) isn't necessarily the whole story :-).


Craig Partridge
craig () bbn com