Until I saw Dean's posting I was also going to just let it go. The trigger
for me was the equating of multi-user license usage to running multiple
copies of software on different systems. Copying software allows more than
one person to execute different functions at exactly the same time. All
DBMS products are engineered to allow us to perform the following
sequential functions:
Accept connection and query
Read state data from cache or disk
Perform operation on all data
Write state data to cache or disk
Respond to user and disconnect
This is the way web servers work and one common way of engineering a
disconnected client/server model. There is no physical way that anyone
process can perform those functions for two users simultaneously, the way
two instances of pirated software does. All users must wait for any user
consuming a license to disconnect before they consume the license in their
own turn. The terms "connect" and "disconnect" may imply logging in or
simply going inactive as someone else consumes the license resource - the
point is that the way MV platforms are engineered, only one user can make
use of the resource at any given moment in time.
This use of licenses has a long standing legal precedent, here are two
examples:
Modems which came into use in 2nd to 3rd generation systems allowed
one user to connect and then disconnect, followed by another user who did
the same. "Per-seat" licensing, compared to "named user" licensing has
always acknowleged this paradigm. Today, if we choose to allow one user to
consume that license for a period of 2 hours before giving it up, or 200
milliseconds, that is a matter of design. There are no set standards for
how long a user must consume a license in order to be considered legal,
except where software vendors impose some minimum connect time. Such
impositions are considered bad design and virtually no company takes a
stance on this because it's technically and politically unreasonable.
Software like UniObjects, PDP.NET, mv.NET, ODBC and RPC, are
specifically designed to allow software to connect, process, and disconnect
as described above. These products and technologies, and many others like
them have established a precedent which allows developers to effectively
consume licenses with as short a connect time as possible. I have not
heard of a single legal case in the IT world, not just the MV market,
brought by a DBMS vendor against a VAR/developer for abuse of this common
communications design. I think any company that does so would be
committing political suicide, if it allowed its developer base to go so
long without action, and then all of a sudden claimed that accepted
practices and established connectivity products were now in violation of
their legal terms.
Unless IBM publicly states their position on this topic, takes a developer
to court, or just sends a polite "please rethink your license consumption"
note to someone, we will not know how liberal they are about their
licensing, regardless of what their license actually says. My guess is
that no DBMS company will take action unless there is blatent abuse. Such
abuse would probably have to involve exploitation of a bug which didn't
enforce licensing policies. I'm also guessing no company would make any
open statements - the "we can sue you if we really want to" ace in the hole
is just enough to keep some people from jumping across that nearly
transparent line between reasonable use and abuse.
I'd also like to get some real numbers on the table. Some people say you
can get 10 users on a single license. Anyone who has implemented a
multi-user environment knows this isn't the case because, as stated above,
the first user will connect and you'll have 9 people waiting in line
afterward. A real high performance remote transaction might take about
250ms, plus or minus depending on network bandwidth, which is still only 4
transactions per second. (Note that's not including common MV operations
like SELECT or EXECUTE.) With 10 users one 1 license you'll have some
users waiting in excess of 2.5 seconds, which is simply unacceptable in
some environments. A more real expectation is 2-4 licenses per 10 users.
In a 100 user shop, that doesn't mean 20 to 40 licenses, it means 25-50
because you'll still need some ports for telnet administration, etc. OK,
maybe 50 licenses for 100 users is too much, but you can easily see that
now we're negotiating numbers far different from a 1-to-10 ratio.
Everyone is looking at this situation as an abuse of licenses. I think we
need to look at licensing from a bigger perspective. DBMS companies don't
make money selling to mom and pop shops. I think IBM and other companies
are happy to know about 100 user sites and 1000 user sites, even if they
only consume 30 to 400 seats for those numbers of users. A 30 license
shop, even for 100 users, is much less hassle than 10 systems with 3 users.
Market share is usually quoted in total numbers of users - never in terms
of some number of named users or other more 1-to-1 perspective. So from a
marketing perspective, the more users a VAR can get on a system the better.
I think this market would be much better off if VARs start selling
non-persistent connections to larger shops, rather than a lot of little
telnet/serial/modem licenses to small shops, but that would involve a shift
in mindset that most MV VARs are not willing to make.
Thanks for the time slice. :)
Tony
Nebula R&D
[EMAIL PROTECTED]
.com
David Jordan david-at-dacono.com.au |U2UG| wrote:
> Hi Dean
>
> What you are doing is not a breach of the license.
> Connect, process, Disconnect and forget is OK.
>
> However if you had persistence in the UvBasic program
> where the past activity of that web user is recorded and
> then used in a future action then you may be crossing the
> line.
>
> What people are doing is through special programs and
> queuing where they can run several users continuously
> with several background processes that stay active. This
> is what IBM has frowned upon.
>
>
> User1 ------- -------
> Process1 \ /
> User2 ------------ Multiplex 1xU2 License -------------
> Process2 / \
> User 3 ------ --------
> Process3
>
> Understand that this equates to a loss of revenue to IBM
> and sooner or later one should expect that they will take
> action.
>
> This is not an ethics issue; it falls under the rules of
> Piracy ie running one Microsoft Word on multiple PCs.
> The criminal and civil actions apply to the customer as
> much as they do to the partner that sells the solution,
> so one has to tread very carefully down this path.
> Worldwide IBM, Microsoft and a number of vendors have
> formed a piracy coalition hiring lawyers and that works
> with Governments and law enforcement to attack piracy.
> Large fines and jail sentences have been given to
> suppliers committing piracy and CIOs' have lost jobs and
> CIOs' and Companies have been fined for breaching license
> agreements. Under SOX this is becoming a bigger issue.
>
> One cannot pretend the rules are not there, there should
> be activities to ensure special licensing arrangements
> are in place for 3rd party applications as they are for
> RedBack.
>
> Regards
>
> David Jordan
> -------
> u2-users mailing list
> [email protected]
> To unsubscribe please visit http://listserver.u2ug.org/
-------
u2-users mailing list
[email protected]
To unsubscribe please visit http://listserver.u2ug.org/
