Conference ulysse::rdb_vms_competition

    thanks for notes file

    Sybase has NO TPC-A numbers, Rdb officially has no TPC-B numbers. Makes
    a comparison real hard. On the other hand, they claim to be an OLTP DB
    vendor but have so far been unable to run the OLTP benchmark.

    We have the leading price/performance around (this week).
    
    Barry

    If you want a proof that TPC-B numbers look much nicer than TPC-A
    numbers just look at the TPC numbers that HP published:
    HP TPC-A 9000-852S Informix Online V4.0  43.3 TPS 23,900 k$/TPS
    HP TPC-B 9000-852S Informix Online V4.0  90.1 TPS 5,300 K$/TPS
    
    This says it all !
    
    Yochai
    

From:	KACIE::JSTRYKER "JAN STRYKER ISB PUBLIC RELATIONS 297-2790"  01-Nov-1991 1051"  1-NOV-1991 11:07:50.27
To:	@HEADLINER
CC:	JSTRYKER
Subj:	HEADLINER: THE SCOOP - SYBASE ANNOUNCES TPC-B RESULTS ON VAX 6600

***************************************************************************
The following information has been sent out over THE SCOOP.  Please forward
this information to others in your area who may need to know. 
***************************************************************************

(forwards deleted)


Subj:	THE SCOOP -- SYBASE ANNOUNCES TPC-B RESULTS ON VAX 6600

[DIGITAL PRESS RELEASE]


        SYBASE ANNOUNCES BREAKTHROUGH TPC-B BENCHMARK RESULTS

       ON VAX 6000 MODEL 600 SERIES:  NARROWS GAP BETWEEN UNIX

                      AND VAX PRICE/PERFORMANCE


Emeryville, CA -- October 30, 1991 -- Sybase, Inc. today announced 
benchmark results of 152.47 transactions per second (tpsB) and 
price/performance of $4,418/tpsB in an audited TPC Benchmark B 
(TPC-B) test of SYBASE SQL Server on Digital's new VAX 6000 Model 
610 system.  These results -- 152.47 tpsB -- represent the highest 
number of TPC-B transactions ever achieved by a single-processor 
computer.  In addition, the SYBASE/VAX 6000 Model 610 combination 
sets a new price/performance record for any VAX computer in a TPC-B 
benchmark.
     The test was conducted on Digital's powerful new VAX 6000 Model 
610 computer running SYBASE SQL Server, Release 4.8.  SYBASE SQL 
Server is a leading client/server-based relational database 
management system (RDBMS) for on-line applications.  SYBASE, Release 
4.8, was designed explicitly for symmetric multiprocessor computers 
such as the VAX 6000 Model 600 series.




     SYBASE SQL Server achieved price/performance comparable to 
RISC/UNIX systems on the new Digital VAX 6000 Model 610 at $4,418 
per tpsB, based on a five-year cost of ownership.  According to 
William R. Demmer, vice president of VAX VMS Systems and Servers, 
Digital Equipment Corporation, "These results reaffirm Digital's 
commitment to deliver leadership price/performance on our VAX VMS 
systems.  We are very pleased with the success of the Digital/Sybase 
TPC-B results and look forward to delivering continued leadership 
performance to our mutual customers."
     Stewart Schuster, Sybase's vice president of marketing, said, 
"The combination of Sybase with Digital's newest VAX series now 
provides price/performance previously perceived to be in the domain 
of UNIX systems. We will continue to work closely with Digital to 
give users leadership performance and price/performance."
     The TPC benchmark suites have been used by many of the major 
hardware and database companies and are considered a consistent 
means of measuring database performance.  Performance is measured in 
an update-intensive database environment characterized by 
significant disk input/out (I/O), moderate system and application 
execution time, and transaction integrity. 
     The TPC-B benchmark also defines rigorous total cost of 
ownership standards for the calculation of price/performance.  Tom 
Sawyer of Performance Metrics audited the TPC-B benchmark.



SYBASE SQL SERVER
     SYBASE SQL Server is a leading client/server-based RDBMS for 
on-line applications, offering customers high performance, 
server-enforced integrity and security, high application 
availability, and open distributed DBMS.  Release 4.8 provides 
significant benefits over other RDBMSs on symmetric multiprocessor 
(SMP) hardware:  high performance, increased user capacity, and 
precise operational control.
VAX 6000
     Digital's new VAX 6000 Model 600 systems now provide leadership 
price/performance, premier functionality and quality, and open VMS 
software along with a complete set of datacenter solutions.  They 
provide two to three times the performance of the VAX 6000 Model 500 
at lower system prices.  They offer continued support of the 
industry's most expandable platform -- the VAX 6000 system's unique 
symmetric multiprocessing platform (SMP) that allows a user to add 
up to five additional processor boards in the same cabinet to 
achieve 6-way SMP with a performance level up to 150 times a 
VAX-11/780.  In addition, these systems use a new CPU chip that is 
the fastest CMOS CPU chip in the industry -- running at 83 MHz.
     Sybase, Inc. develops and markets the SYBASE system, a leading 
client/server-based RDBMS product family for developing on-line 
enterprise applications.  The company offers proven technology 
combined with services and partnerships to provide customers with 
complete desktop to mainframe solutions.  Headquartered in 
Emeryville, CA, Sybase markets its products worldwide through its 
direct sales force, distributors, OEMs, VARs, and system 
integrators.



     Digital Equipment Corporation, headquartered in Maynard, 
Massachusetts, is the leading worldwide supplier of networked 
computer systems, software and services.  Digital pioneered and 
leads the industry in interactive, distributed, and multivendor 
computing.  Digital and its partners deliver the power to use the 
best integrated solutions -- from desktop to data center -- in open 
information environments.
                                ####

Note to Editors:  VAX is a trademark of Digital Equipment 
                  Corporation.

     		  UNIX is a registered trademark of UNIX System 
                  Laboratories, Inc.

     		  SYBASE SQL Server is a trademark of Sybase, Inc.

     		  TPC Benchmark is a trademark of the Transaction 
                  Processing Performance Council.

CORP/92/402

  >>>>>>>>>>>>>>>>>>>>>>    THE SCOOP    <<<<<<<<<<<<<<<<<<<<

The Scoop is a service offered by ISB Marketing Communications to inform
the our field audience ASAP of important news about the VAX 9000 and
related highend topics.  Included in this service are:  Press Articles,
Releases and Quotes, Partylines, Strategy Messages, Analyst Reports and
Conferences, Advertising, etc. 

The Scoop is sent electronically ASAP to the Production Systems Sales
Leaders (PSSL) in the U.S., and our Highend audience in Europe and GIA. 

Please forward this information to others in your area who may need to 
know.

****************************************************************************

	HEADLINER is an on-line method of distributing, quickly,   
		VAX 9000 news to Corporate Digital.  

If you would like to be added to the distribution list, please send mail to: 
KACIE::JSTRYKER.

If you have items that you would like distributed through HEADLINER, please 
contact Jan Stryker at DTN 297-2790.

****************************************************************************

I read a recent Sales Update where Sybase announced 4.8 (Sybase SQL Server),
and it included a 'performance update' where Sybase achieved 261 tps TPC-B
on a 9000-420, 90% of the txns < 0.4 secs response time.

The article stated "Sybase, Inc., and Digital conducted a TPC-B benchmark* of
SYBASE SQL SERVER, release 4.8 running on a VAX 900 model 420. The benchmark
was audited by Codd a& Date, Inc.". So how come we can fork out money to help
Sybase get these results and we can't afford funding for our own Rdb/VMS?
Somethings wrong somewhere...

John.
     

re: .7
�So how come we can fork out money to help
�Sybase get these results and we can't afford funding for our own Rdb/VMS?
�Somethings wrong somewhere...

Somewhere, either in this conference or in IM_PARTNERS, is a discussion about 
whether or not we should spend the money to re-do our tpsB test, since we 
retracted our previous results after the TPC changed its configuration rules at
Oracle's urging.  That rules change made our configuration illegal, after the 
fact.

As far as forking out money for tests, maybe the biggest expense is the people
needed to design and carry out such tests.  A lot of database design and tuning
and application design and tuning is required.  We genereally, I assume, assign
our best peole to do something like that and they are thus not available for 
other things.

In order to help Sybase conduct such tests, all we have to do is provide the
systems and the systems people, PROVIDED BY VMS ENGINEERING, darlings of the
corporation.  This is all speculation on my part.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
But my point is that we must come up with the funding to benchmark our best
hardware (6000-610 and 4000-500) and software Rdb 4.1 and blow the doors
off of everybody else, if we can, or just beat 'em if that's all we can do.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

To quote Herb Edelstein, noted goofball ;-) lecturer on database topics,
"Liars benchmark and benchmarks lie," but we HAVE TO PLAY THE GAME BECAUSE
PERCEPTION IS REALITY.

Bruce

�      <<< Note 1024.8 by MBALDY::LANGSTON "The secret is strong ears." >>>
�                     -< We can and will (must) afford to. >-

�Somewhere, either in this conference or in IM_PARTNERS, is a discussion about 
�whether or not we should spend the money to re-do our tpsB test, since we 
�retracted our previous results after the TPC changed its configuration rules at
�Oracle's urging.  That rules change made our configuration illegal, after the 
�fact.

A slight clarification...

We ran the TPC-B benchmark.  After we had run the benchmark and had the 
numbers approved, the rules for -B were changed by TPC.  Originally, the 
rules were to apply retroactively to previously run -B results.  Digital 
had the choice of retracting our results or re-submitting them using the 
new rules.  At the time we ran the -B benchmark, we used the appropriate 
rules.  As soon as TPC issued the changes, we complied.

We chose to retract the results for various reasons.  Subsequently, TPC 
again changed and made the rule changes effective January 1992.  So, we 
*could* re-submit our figures.  We have chosen not to do this.

�!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
�But my point is that we must come up with the funding to benchmark our best
�hardware (6000-610 and 4000-500) and software Rdb 4.1 and blow the doors
�off of everybody else, if we can, or just beat 'em if that's all we can do.
� !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

This has already been done!  In the October 30 announcement we released 4.1 
TPC-A figures for the 6000-610 and 4000-500.  As I recall, the numbers are 
pretty impressive...  I don't believe that we plan on releasing TPC-B
figures for 4.1 at this time. 

�            <<< Note 1024.9 by COOKIE::OAKEY "NASCAR is racing!" >>>
�                 -< We've got released 4.1 -A numbers already >-

Yes, we have got -A numbers, and I agree they look good - but compared to what?
If your database competitors all publish -B numbers and you publish -A
numbers and say they're the best on VAX, how are our customers supposed to
compare database vendors? Oh, but we say "why don't these other vendors publish
-A numbers - what are they afraid of?". The fact of the matter is that most
of them publish -B numbers and we don't - what impression does that give?

John.

�               <<< Note 1024.10 by HGOVC::DEANGELIS "Momuntai" >>>

�Yes, we have got -A numbers, and I agree they look good - but compared to what?
�If your database competitors all publish -B numbers and you publish -A

John,

Not *all* vendors publish -B...

As of November 1, 

	AT&T/NCR, Bull, Digital, HP, IBM, Sequent, and Unisys 
	publish -A figures thru TPC (Tandem is pending acceptance)

	AT&T/NCR, Compaq, DG, HP, IBM, Oracle, Sequent, Sun, and Sybase
	publish -B figures through TPC (CDC, Digital (Informix-Online, not 
	Rdb), MIPS, Olivetti are pending acceptance)

so slightly more than half publish -B figures.  I don't disagree that it is
a tough sell to convince a customer that -B figures aren't a true
representation of an application when they've already been convinced that
they are. 

But, we are going to have to live with these figures for awhile.  We've got 
impressive -A figures.  Other vendors publish -A figures.  Rdb is an
excellent product with strong features. We've got to work with what we
have... :)  (I know, I know, easy for me to say ;) 

RE: .9
>�!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
>�But my point is that we must come up with the funding to benchmark our best
>�hardware (6000-610 and 4000-500) and software Rdb 4.1 and blow the doors
>�off of everybody else, if we can, or just beat 'em if that's all we can do.
>� !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
>
>This has already been done!  In the October 30 announcement we released 4.1 
>TPC-A figures for the 6000-610 and 4000-500.  As I recall, the numbers are 
>pretty impressive...  I don't believe that we plan on releasing TPC-B
>figures for 4.1 at this time. 

I meant to say "we must come up with the funding..." for TPC-B.

>But, we are going to have to live with these figures for awhile.  We've got 
>impressive -A figures.  Other vendors publish -A figures.  Rdb is an
>excellent product with strong features. We've got to work with what we
>have... :)  (I know, I know, easy for me to say ;) 

What's "a while."  As long as the only certified tpsB numbers available for
Rdb are those produced by Oracle, and are shown side-by-side with numbers for
the ORACLE database that are much, much higher, we're going to have a hard time
with customers wagging ads trumpeting the apparent disparity in our faces.
We need good -B numbers as soon as possible.

I like to ask my customers to ask Oracle why they won't let anyone (except
Oracle) publish performance numbers for the ORACLE database.  I ask this in a
tone insinuating that Oracle's hiding something.  But, in some dimension, our
not having any -B numbers of our own is sort of "the same thing in a different
way."  If the only Rdb TPC-B numbers are those produced by Oracle, some are 
bound to think that they're good enough for a comparison.

Bruce

If you go and read the information posted here, you will see that we can
not publish TPC-B numbers that look good.  So, why publish TPC-B numbers
at all?  This isn't a funding issue (in terms of benchmarking), this is a
benchmark which has been modified to make Rdb/VMS look bad.  It will take
actual engineering before we can run decent TPC-B numbers.

Hal