[Search for users]
[Overall Top Noters]
[List of all Conferences]
[Download this site]
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 |
428.0. "DBMSs on RISC/Ultrix, etc." by VNASWS::EDER (Gottfried Eder, ACT Vienna) Wed Sep 13 1989 16:48
I'd like to share some recent experience and ask you all if you can confirm what
I'm seeing.
A prospect was looking for a DB system, in order to manage large address files
(actually, that's just his business).
He designed a benchmark, which is
- read-only
- large query only
- with sometimes complex queries.
For reasons unknown to me, the only DEC gear that was bid was a RISC machine
(don't ask me which...). They benchmarked Ingres V5 and Oracle V6 on a
DECstation 3100 with 24 MB of memory and one large data disk in addition to the
system disk.
Some results as they were communicated to me at present:
Ingres V5 was almost always way faster than Oracle V6
IBM bid AS/400, and was 10 times slower
HP bid Oracle and was not able to run the tests
NCR bid Oracle, failed, then switched to UNIFY, and failed again - i.e. they
simply were not able to run the tests.
Right now, I'm rerunning much of it on Rdb T3.1 on our 6300, just for the fun
of it.
In most cases where Rdb behaves well, and that's most of them, it's as fast or
faster than our friends on the RISC machine. In other cases, certain things
dont't work quite so well and we're slower.
I'm not at all sure that this can be seen as a real benchmark, I don't know how
well the Oracel and Ingres databases were tuned. I sure did tune the Rdb/VMS
one. But still, the results are interesting, I think, especially given that the
tests are NOT always disk-bound!
Any comments appreciated
Gottfried
T.R | Title | User | Personal Name | Date | Lines |
---|
428.1 | | COOKIE::BERENSON | I'm the NRA | Wed Sep 13 1989 20:37 | 12 |
| One observation is that ORACLE's optimizer sucks. So, a read_only
complex query benchmark is bound to do better with INGRES, even if they
used an older version on INGRES.
It is interesting that we would do well on queries that aren't disk
bound given the 3:1 difference in CPU performance. Well, you do have
faster disks and better controllers on the 6300. So, even there the
comparison is not all fair.
Interesting results however. Given that Digital's database offering is
going to be based on INGRES, it says some good things about our ability
to compete in the Unix market.
|