Re: SUMMARY: Solaris 2.2 licensing

From: Barry Margolin (
Date: Fri Jul 02 1993 - 14:22:43 CDT

In article <20v61m$> (Wolfgang Rupprecht) writes:
[Example of the number of "users" when every workstation mounts the file
systems on every other workstation.]
>Notice that the "users" as sun counts them goes up quadratically with
>the real number of users sitting at keyboards. As anyone involved in
>algorythms can tell you, quadratic growth in the time or cost to do
>something is a *very* bad thing. Things become very expensive to
>scale up.

Note, however, that Sun isn't charging on a per-user basis, so the price
doesn't go up quadratically. They have two license charges: a two-user and
a server price. So the difference between the old and new licensing
schemes is a linear multiple, not a quadratic.

On the other hand, what *does* go up quadratically in such a scheme is
complexity. What I haven't seen anyone mention in this discussion is how
silly such a network architecture is. At our site we try to do as much as
we can with dedicated file servers and dataless workstations. Some of our
compute servers also have local, exported file systems, but none of the
personal machines do. This means that no one hangs when someone reboots
their workstation, and we only have to do backups on a couple dozen
machines rather than hundreds.

Before you complain so much about the linear increase in the license fee,
take a look at the quadratic system administration expense you're giving
yourself! Perhaps Sun is trying to use the license pricing to encourage
customers to use a more reasonable network architecture. :-)

Barry Margolin
System Manager, Thinking Machines Corp. {uunet,harvard}!think!barmar

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:07:59 CDT