| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 714.1 | Silly | POBOX::BOOTH | Bo knows MUMPS | Wed Aug 15 1990 20:08 | 8 |
| As I understand it, those results were obtained with VERY EARLY FT of
V4. The results never should have been published as it is not at all
unusual for early FT performance to be substantially different than
the performance of the product that is released.
---- Michael Booth
|
| 714.2 | Product Manager of Rdb | ROM01::FERRARIS | Discover wildlife... have kids! | Thu Aug 16 1990 15:27 | 73 |
| From the NOVA::RDB_31 notes file:
================================================================================
Note 858.0 BEWARE of VAX4000 Performance Numbers! No replies
NOVA::HORN "Steve Horn, Database Systems" 66 lines 15-AUG-1990 13:48
--------------------------------------------------------------------------------
I've received a lot of mails and phone calls about the performance
numbers recently released for the VAX4000. Here's one of my more
recent replies.
-Steve
From: NOVA::HORN "Database Systems Product Management" 15-AUG-1990 11:42:39.67
To: MAIL::LINCOLN
CC: HORN
Subj: RE: Rdb V3.0 vs V4.0 benchmarks
Myra,
It was unfortunate that the 4000 numbers were released. The 4.0
numbers were using an early field test version and do not represent
actual V4.0 performance. We will re-run the tests when we are
closer to actual shipment, and I can assure you that V4.0 will
out perform V3.1.
Regards,
Steve
From: MAIL::LINCOLN 14-AUG-1990 17:05:01.98
To: NOVA::HORN
CC:
Subj: Rdb V3.0 vs V4.0 benchmarks
I N T E R O F F I C E M E M O R A N D U M
Date: 14-Aug-1990 03:53pm CDT
From: MYRA LINCOLN @STO
LINCOLN.MYRA
Dept: SOFTWARE
Tel No: DTN:445-6563
TO: Remote Addressee ( _NOVA::HORN )
Subject: Rdb V3.0 vs V4.0 benchmarks
Steve,
I recently gave the Database Strategy PID to the Monsanto Rdb Users Group. The
PID talks about a major goal of High Performance. It also discusses some Near
Term features such as Improved Lock Management, Shortened Path Length and
Optimized Commit Sequence - at least some of which are implemented in V4.0. As
Monsanto is particularly concerned with performance, they asked me to follow-up
with a benchmark comparison of Rdb V3.0 and V4.0.
I found such a comparison - using the TPC benchmark on a 4000. However, the
version 4.0 numbers are actually lower than the V3.0 numbers. Is there
something involved that negates the performance improvements that were made to
V4.0? (Such as the overhead associated with 2PC?)
I would like the party line on this issue before I respond to Monsanto's
request.
Thanks!
|
| 714.3 | We need some good numbers. | MBALDY::LANGSTON | Rdb Sales Support Mercenary | Thu Oct 04 1990 02:48 | 26 |
| � We will re-run the tests when we are
�closer to actual shipment, and I can assure you that V4.0 will
�out perform V3.1.
Well... here we are closer to actual shipment (I understand that late
November is somebody's guess). Can we expect to see the
"new-and-improved" results soon? A customer, today, asked me if we
have some benchmark numbers for 4.0. They are looking for a reason to
upgrade. I said "we should have those soon."
I just saw an impressive-looking bound document in our sales
literature room, "TPC Benchmark(TM) A Results for the VAX 4000 Model
300 System usig VAX ACMS, VAX Rdb/VMS, and VAX DECforms Software and
Accompanying KPMG Peat Marwick Attestation Report Number 1 July 1990"
It has a big "DECtp SYSTEMS" sideways (like Bud Dry) on the front
cover.
I'm tempted to go in there after the librarian goes home and
throw them all in the dumpster. In the hands of an unsuspecting
sales or sales support rep, these could be deadly.
It quotes the numbers mentioned in the base note, i.e. 21.7 TPS for
Rdb V3.1 and 21.6 for V4.0.
|
| 714.4 | DO IT | COPCLU::BRUNSGAARD | ACID isn't just music and junk | Thu Oct 04 1990 09:54 | 8 |
| Please do it !!!
The numbers are bogus, and I really hope the person responsible for
these GOOFY-numbers have had his/her share of the noise.
And leave a note to the librarian specifying WHY you have done it.
Lars
|