|
|
|
|
|
|
|
|
|
|
|
|
| home | help | back | first | fref | pref | prev | next | nref | lref | last | post |
Date: Thu, 29 Aug 91 12:33:21 -0400
From: Bob Sutterfield <bob@MorningStar.Com>
To: stan@karazm.math.uh.edu
Cc: craig@sics.se, com-priv@uu.psi.com
In-Reply-To: "Stan Hanks (bcm)"'s message of Thu, 29 Aug 91 10:57:31 CDT <199108291554.AA00553@Menudo.UH.EDU>
Date: Thu, 29 Aug 91 10:57:31 CDT
From: "Stan Hanks (bcm)" <stan@karazm.math.uh.edu>
... dial-up ... only works for low speed applications, like mail
and the odd FTP or so. People in corporateland *usually*
understand that you need at least 56kb to do anything they would
consider useful for commercial applications ... and that T1 may not
even be enough if they're linking multiple sites for the purpose of
doing routine business --
Dialup IP is the hemp of the networking world. Once you've had a
sample, you're intrigued by the possibility of getting more and better
stuff - a bigger, faster buzz.
Some network providers see low-volume, low-speed, low-cost offerings
as a loss leader, something to get the customer into the store. From
the other end of the transaction, the basic infrastructure
applications you note (DNS, NTP, SMTP, NNTP, TELNET, FTP) work just
fine over thin wires. An organization can satisfy its immediate
connectivity needs quickly and inexpensively, and perhaps even throw
away the temporary lifeline when the T1 firehose finally gets
installed. There will be no retraining of users and no rewriting of
applications, and if they were careful when they started, minimal
reconfiguration on the corporate LAN.
Certainly, nobody in h{is,er} right mind will even consider providing
(e.g) a commercial database service over a thin wire. They'll want
the fattest pipe they can buy to the nearest core router. But those
same providers are happy that their consumers can use dialup IP to
gain access to the service, where they had no access before. In that
sense as well, thin wires create a market for a fat pipe.
...take concurrent engineering as an example if you will.
I am familiar with one distributed application currently under
development for a government contract that could not have been
completed if not for the availability of dialup IP. The eventual
deployment will depend partly upon fat pipes, but the engineering
environment consists of developers scattered across the continent,
working concurrently on their own segments of an application based on
TCP sockets carrying simulation data. Since the eventual deployment
will involve time-critical and sensitive information, they even see
the thinwire development environment as having been an advantage: the
design includes more robustness and efficiency than it may have
otherwise.
No, it's not the same sort of thing that can be done over fat pipes,
but it gets the job done.
...there is a large market on the commercial side for usage-based
high speed service...
At the same time, there is also a large market for low-speed service,
with various pricing models. As the top is growing higher, the base
of the networking pyramid is growing lower and wider. And it's eating
up more of the address space faster...
(Disclaimer: Metropolitan Fiber Systems sells fiber connections for
networks; Morning Star Technologies sells PPP for UNIX workstations.
It shouldn't surprise anybody that Stan and I have slightly different
points of view :-)
|
|
|
|
|
|
|
|
|
|
|
|
| home | help | back | first | fref | pref | prev | next | nref | lref | last | post |