| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 1142.1 | Specs have changed | IJSAPL::OLTHOF | Henny Olthof @UTO 838-2021 | Wed Apr 29 1992 09:29 | 9 |
| Jeanne,
Yes, the specs have changed and therefore we withdrew previous TPC-B results.
I cannot beleive that we did the 193.x test with old specs, so I guess the
Oracle rep is trying to confuse the customer. Ask Oracle about thier Rdb test,
which showed 60.1 TPS on the same hardware.
Cheers,
Henny Olthof, TP-DB The Netherlands
|
| 1142.2 | TPC approved release of our V1.0 figures | COOKIE::OAKEY | Picard/Riker '92 | Wed Apr 29 1992 17:31 | 9 |
| � <<< Note 1142.1 by IJSAPL::OLTHOF "Henny Olthof @UTO 838-2021" >>>
� -< Specs have changed >-
Yes, the 193.76 tpsB results were run under V1.0 of the TPC-B benchmark.
However, Digital did not take advantage of any loophole to submit these
figures and get them approved by the TP Council. At the time of submission
to and release by TPC, V1.0 of the benchmark was what was in effect.
|
| 1142.3 | If Oracle is misleading customers we need to know | COOKIE::OAKEY | Picard/Riker '92 | Wed Apr 29 1992 19:22 | 8 |
|
Re: .0
Refer to note 1095 if you feel that Digital's TPC results are being
misrepresented by Oracle. We feel that it is important that Digital's TPC
results be fairly and accurately represented and we need to know if there
are instances when other vendors are presenting misleading information.
|
| 1142.4 | ORACLE's Sour Grapes | COOKIE::BERENSON | Lex mala, lex nulla | Wed Apr 29 1992 20:17 | 37 |
| There are two versions of TPC-B, V1.0 and V1.1. The few TPC-B results we
have published are all V1.0 and were published prior to V1.1 taking
effect (April 1).
At one point we withdrew some older TPC-B V1.0 results (for a VAXcluster)
because a clarification by the TPC required that we prove the
clarification impacted our results by less than 5%. We decided that the
effort to do this wasn't worthwhile because the results were already
obsolete and the cost of re-running the benchmark was high. This whole
area is irrelevant when discussing the recent 6560 results since that
test fully and completely complies with TPC requirements in every way,
shape, and form. ORACLE is just whining because we beat them in a
benchmark. They used every trick in the book to get the TPC to discredit
our results, including forcing us to explain some internal workings of
Rdb that they wanted to claim made it impossible for us to sustain the
tps-rate for 8-hours, but they failed. The results are fully
conformant.
ORACLE gets identical performance for V1.0 and V1.1. Our performance for
the benchmark under V1.1 is unknown, except we know its at least as good
as ORACLE's performance. We see no point in re-running the benchmark since
the TPC considers pre-existing V1.0 results to be perfectly acceptable.
Any future results will have to be done with V1.1, but recall that we are
not very interested in TPC-B. We believe that both TPC-A and TPC-C are
much more representative of what real customers do. We only ran the 6560
TPC-B benchmark to debunk ORACLE's completely misleading advertising, and
had to use the same machine they were using. Further, we see no point in
running another benchmark on a 6560 as it is not a current product
offering.
Tell your customer that ORACLE is just expressing sour grapes over not
being the TPC-B leader. They went out on a limb by advertising their own
benchmark results for Rdb and not applying the same level of application
optimization to the effort as if it had been their own product, and they
had their mistakes shoved down their throat.
Hal
|
| 1142.5 | thanks | VAOU02::JBLAIR | | Wed Apr 29 1992 20:31 | 3 |
| Thank you, everyone for the clarification.
-jeanne
|