[2437] in Commercialization & Privatization of the Internet
Billing systems
daemon@ATHENA.MIT.EDU (drw@BOURBAKI.MIT.EDU)
Thu Feb 27 14:27:11 1992
Date: Thu, 27 Feb 92 14:03:12 EST
From: drw@BOURBAKI.MIT.EDU
To: com-priv@psi.com
In-Reply-To: <3578901@mtsg.ubc.ca> "Mike.Patterson@mtsg.ubc.ca"
From: Mike.Patterson@mtsg.ubc.ca
Increasing the capacity of a link also usually makes the cost of that
link less per kbps than a lower speed link -- telecommunications are
characterized by substantial economies of scale.
This works until you saturate the technology available. I suspect
that (before fiber optics) if you multiplied the long-distance traffic
level between N.Y. and D.C. by 10, ATT would have to install 10 times
the copper that it had before. There's not much economy of scale
there. What happens when the backbone traffic would fill 100 T-3's?
From: Craig Partridge <craig@aland.bbn.com>
The proportion of traffic leaving an organization's
LAN into the greater Internet has proved to be almost constant, across
all organizations.
But this is more a property of the types of organizations that are now
attached to the Internet (universities) than of networking in general.
Consider an individual, whose network activity is to send and receive
10 pieces of e-mail a day. Or better (as someone else noted) a
company with dozens or hundreds of field operations, each of which
needs to use the network very intensively for only a few minutes each
day. In both cases, the peak/average ratio is quite high. The only
case where usage is nicely constant is where each site is large enough
to *have* a LAN.
Dale Worley Dept. of Math., MIT drw@math.mit.edu
--
Newsgroups = ASCII television -- Alex Lopez-Ortiz
