Re: Db2, Oracle, SQL Server

"DA Morgan" <damorgan_at_x.washington.edu> wrote in message news:1108004465.797766_at_yasure...
>
> > That is very "convenient" of you to claim (earlier in this thread) that
DB2
> > cannot scale to large workloads in OLTP, and then claim the TPC-C
benchmark
> > is irrelevant.
>
> You mean this is mutually exclusive? Try a bit of Boolean logic for a
> change. It is possible for two different things to be true.
>
> 1. DB2 doesn't scale large OLTP workloads = TRUE
> 2. TPC benchmarks are irrelevant = TRUE
>
> Which is not to say that I agree with #1 but 1 and 2 are not mutually
> exclusive. You must write fascianting IF/THEN statements.
> --
> Daniel A. Morgan

Daniel constantly berates the TPC-C benchmarks as a measure of OLTP database performance, but he has never supplied any legitimate reasons as to why it is irrelevant.

Does Daniel mean that the TPC benchmark is fraudulent and does not reasonably measure OLTP performance? Or does he merely say that performance is irrelevant because there is not a big difference among databases and people don't decide on which product to use on that basis?

Daniel also says that it is completely irrelevant when judging the validity of the TPC benchmarks as to whether Oracle supports the TPC and whether they participate heavily in the TPC by supplying staff to the committees (since its inception more than 15 years ago). If Oracle is such a strong supporter of TPC, I don't believe one can (whenever convenient) brush the TPC aside and call it irrelevant if one is trying to claim that DB2 cannot scale on OLTP like Oracle.

One might (at your own risk) say that performance is irrelevant, and therefore the TPC is irrelevant, but that is not what was said in this thread. The statement was made by a previous poster that DB2 does not scale on OLTP. That is a statement about performance..

That statement about DB2 not being able to scale was made even though DB2 does in fact scale 2.5 times more on the TPC-C benchmark than the fastest Oracle TPC-C that has been published. So if someone thinks that DB2 cannot scale on OLTP, and they don't think TPC-C is a legitimate benchmark for OLTP performance, then I would like to see another benchmark that demonstrates both of these things.

Basically it comes down to this:

The claim (and subsequent defense of that claim) that DB2 does not scale on OLTP is absurd. DB2 total throughput measured as transactions per minute (TPM) on TPC-C is 2.5 times more than the fastest Oracle TPC benchmark published using RAC and same number of processors (64). The DB2 cost per TPM is cheaper than the Oracle benchmark cost per TPM. http://www.tpc.org/tpcc/results/tpcc_perf_results.asp

So Daniel Morgan cries ABSURD at the TPC, like a little boy who kicks the rock he has just stumbled on. Received on Wed Feb 09 2005 - 23:27:22 CST