Re: lies damn lies and benchmarks

"Chris Weiss" <chris_at_hpdbe.com> wrote in message news:<aauvlh$2200$1_at_msunews.cl.msu.edu>...
> I live eat and breath Oracle most days, but unfortunately on equivalent
> hardware, Oracle gets creamed by SQL Server:
>
> http://www.tpc.org/
>
> Even on larger Unix boxes, smaller Windows machines with SQL Server tend to
> outperform Oracle. Of course, this says nothing about stability, cost of
> ownership, etc. In my experience, it is *much* easier to administer and
> tune a stand alone SQL Server machine compared to a stand alone Oracle
> database on Unix, but with multiple boxes requiring several similar
> databases, Oracle is much easier to administer. To Oracle's credit, I have
> yet to see a BSOD or its equivalent where there was not a serious hardware
> problem on a high end Unix box running Oracle, but Oracle does and has had
> its fair share of bugs.
>
> In the long run, the choice of database really comes down to what the client
> wants, what the needs of the application are, and what the host site can
> afford. Oracle has a much higher dollar to performance ratio than SQL
> Server, which is why SQL Server is still around and Oracle still needs to
> stay hungry and competitive.
>
> Since I like administering things in Unix so much better than Windows, I
> will always have a bias toward software that runs on Unix. However, that is
> no reason not to be honest about my bread and butter - Oracle.
>
> I think DB2 and SQL Server both pose serious threats to Oracle, and I hope
> that Oracle takes this seriously and continues improving its product. From
> my perspective, Oracle has dropped the ball with respect to performance in
> favor of features. Since DB2 and SQL server outperform Oracle and are
> catching up with respect to features, this can only go on for so long before
> Oracle begins to fall back quickly.
>
> --
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> Chris Weiss
> mailto:chris_at_hpdbe.com
> www.hpdbe.com
> High Performance Database Engineering
> Available for long and short term contracts
>
>
> "Niall Litchfield" <n-litchfield_at_audit-commission.gov.uk> wrote in message
> news:3cd14594$0$232$ed9e5944_at_reading.news.pipex.net...
> > I see that according to the high performance labs at accupuncture (sorry
> > accenture) that MSSQL outperforms Oracle running SAP R/3. It turns out
> that
> > this is MSSQL 2000 sp2 compared with Oracle 8.0.4! Does anyone know how
> old
> > 8.0.4 actually is?
> >
> >
> >
> > --
> > Niall Litchfield
> > Oracle DBA
> > Audit Commission UK
> > *****************************************
> > Please include version and platform
> > and SQL where applicable
> > It makes life easier and increases the
> > likelihood of a good answer
> >
> > ******************************************
> >
> >

Read the complete benchmark document instead of just the results. This was accomplished with a 32 node federated database each with 4 cpus plus a couple of transaction monitors at 4 cpus each. All on NT...I wouldn't want to maintain it. Oracle achieved 50% of the benchmark on less than half the number of CPUs with slower CPUs at that. If you carry through Oracles clustering and apply current clustering multipliers Oracle would achieve a TPCC of over a million with a similar number of CPUs at the same speed.

Do your homework or stay home.

Mike Received on Mon May 06 2002 - 09:17:22 CDT