More details on the NT Workstation TCP limit

Rohit Khare ((no email))
Wed, 7 Aug 96 13:03:06 -0400

This is Tim O'Reilly's second-generation complaint.
He is correct, the new license is MORE restrictive. It also calls into
question multiplexing services, like HTTP-NG which could offer
N "connections" over one TCP socket.



>From Wed Aug 7 12:27:49 1996
Date: Wed, 7 Aug 1996 12:13:22 -0400 (EDT)
From: "Donald E. Eastlake 3rd" <>
Subject: FWD: MS NT Workstation 4.0 Maintaining Limitations

[Deliberately crippling products to manipulate the marketplace is a
strong sign of monopoly power... dee3]

Subject: MS NT Workstation 4.0 Maintaining Limitations

Less than two weeks ago, I expressed my deep concerns to
Microsoft about their proposed limits on the number of sockets
in NT Workstation 4.0. Although Microsoft has publicly backed
down from their plan to build the limitation into the software,
the most recent license for the product keeps such restrictions in the
license--and even expands them. Microsoft's public reversal appears merely to be a strategic retreat.

Here's the wording of the license sent out with NT Workstation
RC2 (Beta B):

" may permit a maximum of ten (10)
computers to connect to the Workstation
Computer to access and use services of the
SOFTWARE PRODUCT, such as file and print services
and peer web services. The ten connection
maximum includes any indirect connections made
through software or hardware which pools or
aggregates connections."

That means that the limitation has been expanded, from "10
users in 10 minutes" (the original limitation) to "10 users
(period)." We believe that Microsoft's position amounts to nothing
more than a "land grab" in the uncharted territory of the Internet.

While at first blush it might seem logical that Microsoft
has the right to set the licensing terms for their own
products, and to make reasonable distinctions between NT
Workstation and NT Server, I believe that in this license,
Microsoft is taking the further step of limiting the use of
the TCP/IP protocol for their users. TCP/IP is not a Microsoft
product, and I don't believe Microsoft has the right to tell
application vendors and users what they can and can't do with it.
TCP/IP is a fundamental service for internetworked systems.

If you accept that Microsoft has the right to tell users
how many sockets their applications can have open, you
must also accept that they have the right to tell users
how much memory their applications can use, or how much
processing power.

As I've pointed out in my letters to Microsoft, at bottom, I
don't want to argue on the basis of whether it's legal or even
moral for them to try to use their control over the operating
system to freeze out competing application vendors. Instead, I
want to argue that what they propose is bad for the Internet.

Because Microsoft is in the unique position of controlling
the operating system as well as competing in the application
space, they have a special responsibility to use that
control wisely.

Microsoft argues that they simply want to position NT
Workstation as a desktop operating system, and that if users
want to run servers, they should use NT Server. Such a view is
short sighted, since it presupposes that we already know what
users want and what developers can create.

Consider the following analogy: If IBM had had the control over
the operating system that Microsoft now has, one could imagine
them saying, back in the mid-1980's:

"When we created the IBM PC, we never meant that users
should do so much on the desktop! This is hurting our
mainframe revenue. So tell you what, we'll give you a
special set of tools for the desktop that will let you
create small spreadsheets and databases there, but if you
want to do any serious computing, you have to use a

While this analogy is a bit farfetched (mainly because IBM
didn't hold all the cards in the way Microsoft does!), a few
problems are obvious. Such a move would have choked off the
waves of innovation that made up the PC revolution. IBM
couldn't imagine then how much people would do on the desktop.
I maintain that Microsoft can't imagine now how much people will
do with the Web on the desktop. When you build in limits from
the start, you get what you build...limits. We are still only
at the beginning of the web revolution, and we *must* keep the
system open, for the applications that have not yet been
imagined or invented.

A final point: Microsoft's attempt to get the Internet
community to accept via a license agreement a limitation that
they clearly found repugnant when encoded in the software seems
like a dangerous trojan horse offering. If users accept the
license now, what is to stop Microsoft from coming back six
months or a year from now and setting limits in the software.
After all, by then they could say: "We're just enforcing it
now, that limit has been there for a long time--since NT
Workstation 4.0!"

The Internet community understood the implications of the
technical limitation and forced Microsoft to back down.
Microsoft stated publicly in their July 19 press release that
"rigorous customer beta testing and subsequent customer feedback
...led to this decision [to remove the limits]." Let's make
them stand by that statement, and remove the limits from their
license as well as from their code.

To reiterate my basic point: Microsoft didn't invent the
Internet or the TCP/IP protocols. They came late to a great
party. They have a choice: they can join the party or they can
try to shut it down. If they try to shut it down, they can
expect that the rest of the people there are going to complain.
And at this particular party, the Internet gives the partygoers
a pretty loud megaphone.

Tim O'Reilly @ O'Reilly & Associates, Inc.  Publishers of Nutshell Handbooks
103 Morris Street, Sebastopol, CA 95472
707-829-0515 ext 266, Fax 707-829-0104,
Check out,,