SunMathematica: the CPU is the computer

[Steve C. Simmons]

Kemp at dockmaster.ncsc.mil (Dave Kemp) writes:

>For a company that advertises the advantages of networking, licensing
>software on other than a network basis (a la FrameMaker, Publisher, etc)
>strikes me as disengenuous.  I'm considering returning the software sight
>unseen and just using it on one of the Macs, which at our site do not sit
>on people's desks.

I strongly suggest you do it, but please inform both Sun and the
Mathematica folks *why* you are doing it *in writing*.  A reasonably
secure (ie, as secure as most PC copy protection) floating licence server
is implementable in a few weeks (I know, I did it).  But a real licence
server takes one hell of a lot more than that.

Consider some of the needs of floating licences.  Let's say you bought
four licences at $10,000 each for a CAD/CAM package.  You find your
engineers cannot use it because a secretary is doing his homework on it.
Should your licence server be able to:

  1. Report on who is using the product?
  2. Restrict usage to given hosts?
  3. Restrict usage to given users?

Take some other possibilities.  User A is running the package.  His
machine burns up.  How do you free up the licence usage?  If you let the
customer do it on site, he can use that capability to break your
licencing.  There are some fairly decent solutions, but they require
fairly hefty (intrusive) mods to the product source.  What's your choice,
modify the product or leave the hole?

Now you get into a classic proprietary/open problem.  If Sun implements a
licence server, they are effectively dictating a solution to all Sun
customers and software vendors.  If this solution conflicts with other
vendors (DEC, Apollo, etc) you've got problems in a heterogenous network.
If the software vendor does his own, as Frame did, your customers now have
to deal with multiple licencing schemes and have lost the capability of
interacting with vendor-supplied solutions.

Floating licences sound wonderful.  We know they *can* be done because
they *have* been done -- there are at least three independant
implementations (Frame, Apollo, Schlumberger Technologies).  But none is
really satisfactory, and none were easy.  If you are in Mathematicas
shoes, where do you put your effort: into improving the graphical
interface to SunView, which will radically increase your sales/income, or
the licence server, which will leave it roughly the same?

 
Steve Simmons		          scs at vax3.iti.org
Industrial Technology Institute     Ann Arbor, MI.
"Velveeta -- the Spam of Cheeses!" -- Uncle Bonsai