| 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 |
Hello, We need some urgent information on Rdb and ACMS. For a customer offer we need to give some performance test results. I thought of giving debit/credit benchmark results, but : 1) I need an exact description of what the debit/credit test is doing 2) I need externally verified results that can be published or were allready published externally. We need results for a 63XX and a 38/9XX VAX preferably, but not necessary not using front end machines. We would we most grateful to get something before the weekend as the (lots of $$$ the next years) offer needs to get out by monday. Thanks to everybody helping a hand. Dominique Dierick
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 350.1 | WELSWS::GIBSON | Alan Gibson @WLO | Mon May 15 1989 13:28 | 4 | |
Sorry this may be too late for you...
Your local Rdb product manager (in the UK, Mike Hudgell) should
have a copy of the audited Credit/Debit benchmark.
| |||||
| 350.2 | Benchmark Numbers.... | HOTAIR::DAVIS | Gil Davis @AQO DTN 554-7245 | Mon May 15 1989 17:46 | 89 |
This isn't everything asked for in .0, but it's a start..
What's really needed is a copy of the TP Benchmark report(s).
From: LINDA::WRIGHT "TP Systems Performance Analysis 10-Apr-1989 1606" 10-APR-1989 16:37
Subj: Summary of DebitCredit Benchmark Results
Below is a summary of the DebitCredit tests Digital has run, internally
audited, and externally published. Tests to which an external auditor
has attested, are indicated by an asterisk (*) following the processor.
Digital Relational Systems
--------------------------
CPU MEMORY DISK DISK DK-CTL 5 YR 95%/1sec 5 YR
SOFTWARE MODEL MB GB ARMS Type/# COSTS TPS K$/TPS
------------------- ------ ------ ---- ---- -------- ------- -------- ------
ACMS 3.0/Rdb 3.0 3400* 20 1.2 6 EDA640/1 $142K 3.5 $41
ACMS 3.0/Rdb 3.0 3600* 32 1.8 4 KDA50/1 $224K 5.7 $39
ACMS 3.0/Rdb 3.0 3800* 64 2.4 6 KFQSA/1 $284K 7.7 $37
ACMS 3.0/Rdb 3.0 3900* 64 3.0 4 KDA50/1 $299K 7.8 $38
ACMS 3.0/Rdb 3.0 6210 64 2.5 4 KDB50/1 $436K 6.0 $73
ACMS 3.0/Rdb 3.0 6220* 64 5.6 9 KDB50/3 $697K 11.0 $63
ACMS 3.0/Rdb 3.0 6240* 128 6.2 10 KDB50/3 $1,006K 18.2 $55
ACMS 3.0/Rdb 3.0 8810 128 6.2 10 KDB50/3 $1,137K 12.0 $95
ACMS 3.0/Rdb 3.0 8820 128 6.8 11 KDB50/3 $1,463K 20.8 $70
ACMS 3.0/Rdb 3.0 8830 128 7.5 12 KDB50/3 $1,804K 27.5 $66
ACMS 3.0/Rdb 3.0 6310 64 4.4 7 KDB50/2 $535K 8.4 $64
ACMS 3.0/Rdb 3.0 6320* 64 6.8 11 KDB50/3 $778K 16.0 $49
ACMS 3.0/Rdb 3.0 6340* 128 9.3 15 KDB50/4 $1,183K 25.4 $47
ACMS 3.0/Rdb 3.0 6360* 192 11.8 19 HSC70/2 $1,855K 30.0 $62
Digital Flat File Systems
-------------------------
CPU MEMORY DISK DISK DK-CTL 5 YR 95%/1sec 5 YR
SOFTWARE MODEL MB GB ARMS Type/# COSTS TPS K$/TPS
------------------- ------ ------ ---- ---- -------- ------- ------- ------
DECintact 1.0A/Hash 3400* 52 2.2 6 EDA640/1 $203K 8.6 $24
DECintact 1.0A/Hash 3600* 32 1.8 4 KDA50/1 $222K 6.2 $36
DECintact 1.0A/Hash 3900* 64 4.3 7 KDA50/2 $420K 12.0 $35
DECintact 1.0A/Hash 6210* 64 5.0 8 KDB50/2 $526K 12.0 $44
DECintact 1.0A/Hash 8810* 256 8.7 14 KDB50/4 $1,360K 26.2 $52
DECintact 1.0A/Hash 2x8810* 512 15.0 24 HSC70/2 $2,853K 47.5 $60
IBM Relational Systems
----------------------
CPU MEMORY DISK 5 YR 95%/1sec 5 YR
SOFTWARE MODEL MB ARMS COSTS TPS K$/TPS
-------------------- ---------- -------- ---- ------- -------- ------
VSE/CICS/SQL/DS 9377-90* 16 10 $705K 5.2 $136
MVS/XA CICS/DB2 4381-Q13 24 8 $1,688K 4.0 $422
MVS/XA CICS/DB2 3090-200E 128+64 32 $8,698K 38.0 $229
IBM Flat File Systems
---------------------
CPU MEMORY DISK 5 YR 95%/1sec 5 YR
SOFTWARE MODEL MB ARMS COSTS TPS K$/TPS
-------------------- ---------- -------- ---- ------- -------- ------
VSE/CICS/VSAM 9377-90* 16 9 $695K 6.0 $116
VM/VSE/CICS/VSAM 9377-90 16 8 $710K 2.0 $355
| |||||
| 350.3 | Some D/C basics.... | HOTAIR::DAVIS | Gil Davis @AQO DTN 554-7245 | Mon May 15 1989 17:50 | 18 |
From the Benchmark report, the Pseudo-code reads as follows:
Begin-transaction
READ MESSAGE FROM TERMINAL (100 bytes)
REWRITE ACCOUNT (random)
WRITE HISTORY (sequential
REWRITE TELLER (random)
REWRITE BRANCH (random)
WRITE MESSAGE TO TERMINAL (200 bytes)
Commit-transaction
Data Types required:
Branch Records containing 100 bytes each.
Teller records " 100 " "
Account " " 100 " "
A 90-day history file containing 50 bytes per entry
| |||||
| 350.4 | Volumes for D/C | HOTAIR::DAVIS | Gil Davis @AQO DTN 554-7245 | Mon May 15 1989 17:56 | 20 |
Some volumes of data.....
The number of records the history area must accomodate can be
determined as follows:
rec sec hr
1 --- * 3600 --- * 8 --- * 90 days = 2,592,000 records
sec hr day
For an X TPS system, the database would be sized as follows:
10X Branch Records
100X Teller Records
100,000X Account Records
2,592,000X History Records
where: X is the TPS load for the system
| |||||
| 350.5 | thanks | KETJE::DIERICK | BLACK HOLES ARE OUTASIGHT | Wed May 17 1989 18:34 | 5 |
Thanks for all the replies, I collected the necessary information. Dominique | |||||