Patrick wrote:

     >I've been to the Transaction Processing Performance Council's web
     >site, <http://www.tpc.org> but have been unable to find any TPC
     >benchmarks for MPE systems (there are several listed for HP-UX). I
     >seem to remember that HP had done some benchmarks at one time, but
     >they were probably TPC-A or -B (? - need to upgrade my memory, I
     >guess :-). Does anyone know if HP has done TPC-C or TPC-D
     >benchmarking on MPE and whether the results were published? (And, if
     >not, why not?)

     To the best of my knowledge, it has been some time since CSY did any
     TPC benchmarks.  First, it is very expensive(time and dollars) to do a
     benchmark, especially one like TPC since there are many rules to
     follow and it must be independently audited.  That expense must be
     weighed against the benefit of doing the benchmark and, in my opinion,
     there are not enough benefits to justify the cost.  Why do I say that?
     I do not like to pigeonhole or stereotype people or companies but, in
     the world of HP3000 sales, we are not usually dealing with people who
     are comparing the performance of the HP3000 against another platform.
     And isn't that the primary purpose of an "industry standard"
     benchmark?  We are usually (1)dealing with customers who already have
     an HP3000 and are looking to add or upgrade.  In that case, since they
     are familiar with how their application runs on their HP3000, a
     comparison of their HP3000 to other HP3000s provides them with a
     better gauge than an "industry standard" benchmark which is running
     code and doing things that they are probably not.  Or, (2) we are
     dealing with a customer who does not have an HP3000, but is looking at
     the HP3000 because that's where the application runs.  That is, the
     application is driving the decision to purchase the HP3000.  When the
     applications(those that are causing customers today to buy HP3000s)
     become readily available on other platforms, then there may be a need
     to have something like TPC benchmarks in order to compete.  In my
     current experience, the applications that are bringing new customers
     to the HP3000 only run on the HP3000.  If I am wrong on that, please
     do not hesitate to let me know.

     If a customer has not picked an application, and is considering
     whether or not to look for one that runs on the HP3000, I will try to
     provide him with reliability, high availability, and supportability
     information to encourage a look at HP3000 applications.  Again,
     performance comparisons do not generally come into play at that point.

     So, that's my opinion.  I will be happy to hear any points on why CSY
     should go back to doing TPC (or any other industry standard)
     benchmarks.  If there are enough benefits to it, then maybe they'll
     start again.

     Jon Broz
     HP Cleveland

     "Definitely my opinions... not necessarily HP's"