| 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 |
C/A (Cullinet) just tested their database on Litel's 9210 and got lower
than expected transaction rates (compared to rates on 6000-class
machines). Obviously we're not very happy with that, and part of the
concern stems from a rumor that the DBA here heard that both Oracle and
RDB got poor transaction rates until they "did something" and then it
really took off. Has anyone heard this or perhaps have an idea what
"something" is? The benchmark is completely CPU-oriented, as they have
only a 10 meg database and 25 meg of buffers. There's the initial I/Os
to load it, and then they drop off to zero and just crank in memory.
Journalling was being done to an ESE20, and they didn't include writing
of "dirty" buffers back to disk in their statistics (for some reason).
Thanks in advance
Bill Valentine-Cooper
(cross posted in VAX9000_PERFORMANCE)
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 711.1 | here's my thoughts for your $.01 | DPDMAI::DAVISGB | Father of three ninjas and a ninjette | Mon Aug 13 1990 22:03 | 8 |
Is this a batch-test we're describing? I've seen Rdb beat the brains
out of Sybase and Ingres (both were dong bulk data loading) by just
committing 100 trx at a time vs one at a time....
Call the TP performance group (Bill Zahavi, et al...) and see what they
think....
Gil
| |||||
| 711.2 | You want speed? | DSM::CRAIG | Nice computers don't go down :-) | Wed Aug 22 1990 04:01 | 2 |
And when you call them ask them about the results of TPC-A using VAX
DSM...
| |||||