[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference ulysse::rdb_vms_competition

Title:DEC Rdb against the World
Moderator:HERON::GODFRIND
Created:Fri Jun 12 1987
Last Modified:Thu Feb 23 1995
Last Successful Update:Fri Jun 06 1997
Number of topics:1348
Total number of notes:5438

1091.0. "DSM is four times faster tahn Rdb" by ZPOVC::JEFFREYCHOY (What's Life After DEC ??) Thu Feb 20 1992 05:13

    I was told by someone that a telephone directory database using
    VAX/DSM is 4 times faster than Rdb/VMS. The benchmark is done sometime
    in DEC 91.
    
    I know that DSM is excellent in complex query kind of application
    and they uses some kind of B-TREE structures.
    
    I need to confirm that database on DSM for query nature kind of
    application can outperformed Rdb Version4 by 4 times regardless of
    database size.
    
    Any info is most appreciated and welcome.
    
    
    Thanks, JC
T.RTitleUserPersonal
Name		DateLines
1091.12 times faster with TPC-ASUOSW3::KAISERThink twice...Thu Feb 20 1992 10:384
Our TPC-A benchmarks show that DSM is two times faster than an equivalent 
ACMS/Rdb configuration. The latest comparison I have was from Sept. 1990.

-Hans
1091.2no more32466::NEEDLEMANtoday nas/is, tomorrow...Thu Feb 20 1992 14:356
    if you used rdb 4.0 numbers, this might have been true. if you look at
    the new numbers with rdb 4.1 this is VERY different.
    
    Barry
1091.3Numbers, please?SUOSW3::KAISERThink twice...Thu Feb 20 1992 14:560
1091.5Latest TPC-A resultsCOOKIE::OAKEYThe Last Bugcheck - The SequelThu Feb 20 1992 17:0822
�             <<< Note 1091.3 by SUOSW3::KAISER "Think twice..." >>>
�                             -< Numbers, please? >-

Otay...

TPC-A local results as of Feb 7, 1992.

VAX 4000-300	VAX DSM V6.0 (VMS V5.4)		41.40 tpsA
		VAX Rdb/VMS V4.1 (VMS V5.5)	31.90 tpsA

VAX 6000-510 	VAX DSM V6.0 (VMS V5.4)		68.50 tpsA
    (DSSI) 	VAX Rdb/VMS V4.0 (VMS V5.4)	33.90 tpsA

VAX 9000-210	VAX DSM V6.0 (VMS V5.4)		143.4 tpsA
		VAX Rdb/VMS V4.0 (VMS V5.4)	69.40 tpsA


There are other V4.1 figures, but the 4000-300 is the only official TPC-A 
benchmark where both Rdb and DSM have TPC accepted results.

All Rdb TPC-A benchmarks use ACMS and DECforms as part of the benchmark.
1091.6These aren't official as far as TPC is concerned...COOKIE::OAKEYThe Last Bugcheck - The SequelThu Feb 20 1992 17:1213
�      <<< Note 1091.4 by NOVA::FEENAN "Jay Feenan, Rdb/VMS engineering" >>>
�                         -< off the top of my head... >-

�    6630 - 220TPS
�    6640 - 275
    
Jay,

Are you sure these are correct for TPC-A?

My understanding was that we really haven't done a 6630 TPC benchmark and
the 6640 number sounds like a TP1 result.
1091.7seek and ye shall findDATABS::NEEDLEMANtoday nas/is, tomorrow...Thu Feb 20 1992 18:298
    Actual TPC submitted numbers may be found at
    
    TPSDOC::SYS$PUBLIC:[TP_PERFORMANCE]TPC_SUMMARY.TXT
    and
    TPSDOC::SYS$PUBLIC:[TP_PERFORMANCE]TPC_TPC_COMPARISON.TXT
    
    
    Barry
1091.8NOVA::FEENANJay Feenan, Rdb/VMS engineeringThu Feb 20 1992 20:299
    re:-.1
    
    Yup your correct these are TP1 numbers...If I get time, I should look
    through my mail to figure out the tp1 comparison between 4.0 and 4.1
    since these would reflect the database part of the performance
    increase.
    
    -jay
1091.9No DECformsTPSYS::SHAHAmitabh Shah - Just say NO to decaf.Fri Feb 21 1992 19:3216
> COOKIE::OAKEY in .5

> All Rdb TPC-A benchmarks use ACMS and DECforms as part of the benchmark.

No, NONE of the Rdb 4.1 TPC-A benchmarks use DECforms (although some previous
results with Rdb 4.0 did use DECforms). 

I think the most significant comparison is the 4000-300 test, since it compares
Rdb 4.1 with DSM. Those with Rdb 4.0 don't count anymore. 

> Re. 6640 275 TPS (.4) 

There are tests underway currently, but we don't have any TPC-A numbers for
any 6600 class SMP machine. 275 TPS for 6640 certainly looks very optimistic.

-amitabh.
1091.10NOVA::FEENANJay Feenan, Rdb/VMS engineeringFri Feb 21 1992 20:3911
*> Re. 6640 275 TPS (.4) 
*
*There are tests underway currently, but we don't have any TPC-A numbers for
*any 6600 class SMP machine. 275 TPS for 6640 certainly looks very optimistic.

These were pertaining to some TP1 numbers that were generated, my 
mistake.  I've deleted my reply so that people wouldn't get confused 
with the numbers relative to TPC-A

-Jay
1091.11What forms then?HGOVC::DEANGELISMomuntaiSun Feb 23 1992 05:029
�    <<< Note 1091.9 by TPSYS::SHAH "Amitabh Shah - Just say NO to decaf." >>>
�                                -< No DECforms >-

�No, NONE of the Rdb 4.1 TPC-A benchmarks use DECforms (although some previous
�results with Rdb 4.0 did use DECforms). 

??Then what is being used as the forms component? Surely not TDMS.

John.
1091.12No need to use a forms package.TPSYS::SHAHAmitabh Shah - Just say NO to decaf.Mon Feb 24 1992 17:469
	Re. .11
	
	> ??Then what is being used as the forms component? Surely not TDMS.
	
	The forms requirements in TPC-A are trivial enough to write a small
	user agent that does the actual work. This improved the performance
	quite a bit, since DECforms are very CPU intensive. 

	-amitabh