Patrick suggests:
> Has anyone sent a message to Carly F. that a benchmark team is a resource
> that *needs* to be added at CSY?

CSY's usual excuse is that they don't think that it is worth committing the
resources to, at the cost of leaving other things not done.

This is essentially a correct statement, IMHO, as such a project should
*not* be a CSY engineering project, it should be a CSY *marketing* project.
Marketing should be paying for this kind of thing, and should arrange for
both the funding and the personnel to carry it out.  It's up to marketing to
decide if having great TPC numbers for the 3000 would be a more valuable
sales tool than, say, a month of magazine advertising, etc.

I'm skeptical that a "shared source" project would get anywhere, but then
again this is an "application" programming project, not some esoteric
technical thing, so there ought to be a wider group of potential
participants for this than for, say, Query enhancements.

Lots of people consistently miss a key point about open/shared source, which
is that it only works if there is someone willing to spend virtually their
full time driving and managing the project.  Look at all of the *successful*
open source projects and at the core you'll find one individual who drives
the project and who did all of the original work.  That central person also
usually has to complete at least a basic implementation before others can be
interested in the project.  You can't make stone soup without a stone.  So
open source usually won't work to get other people to do the things that
*you* think should be done.

Having said that, the best way to do TPC-C on the 3000 would probably be
with a group effort that included input from all of the most experienced
application and database designers and performance experts on the 3000.  If
CSY Marketing would provide the funding for the one technical project
coordinator/architect to oversee, manage, and drive the project, or if
someone in the community is willing to voluntarily take on this role, then
there would probably be lots of people willing to offer advice and possibly
programming effort to the project.

Another option would be for one 3rd party company who depends on the 3000
for their existence, or a consortium of such companies, to provide the
funding required to get the project going.

Again, the first step obviously should be to try to determine just how much
work would be involved, and what the projected results would be.  If we
can't get TPC-C numbers that are substantially better than the current HP-UX
numbers for equivalent hardware then there's no big win in doing it.  Of
course if we *can* blow the HP-UX numbers away, then there's potentially a
political problem inside HP of showing up the HP-UX side of the house.

G.