[NTLUG:Discuss] Stress testing an RDBMS

[Christopher Browne]

On Sat, 12 Aug 2000 14:13:52 CDT, the world broke into rejoicing as
"Alex D. Coker" <adcoker at leap4linux.com>  said:
> This is a multi-part message in MIME format.
> --------------5AAA9CB142B4B79DB09F3DFA
> Content-Type: text/plain; charset=us-ascii
> Content-Transfer-Encoding: 7bit
> 
> I've received an evaluation of the Empress RDBMS for the NetWinder
> internet appliance that I presented at  July's meeting. Can someone
> assist by describing a good stress test for this RDBMS?
> 
> I have the system software loaded and operational. I can test the
> programming pieces just fine, but the RDBMS itself is another story. I
> can create tables and etc. but I'll need some example things to do in
> the DB itself. Should I setup multiple triggers and stored procedured to
> simulate a real world performance test?

a) TPC-C.
   <http://www.tpc.org/cspec.html>

b) Wisconsin Benchmark

c) See The Benchmark Handbook: For Database and Transaction Processing
   Systems, Edited by Jim Gray [then of DEC, now of Microsoft]
   <http://www.mkp.com/books_catalog/2925toc.htm>

Making these into "real" tests is quite a lot of work, and often involves
pulling in consultants from _all_ the vendors involved.  [e.g. - for
a benchmark of Oracle on Solaris with Tuxedo, there would probably be
people from Oracle, Sun, and BEA in to do tuning work]
--
cbbrowne at ntlug.org - <http://www.hex.net/~cbbrowne/linux.html>
"What's wrong with 3rd party tools? Especially if they are free?  What
the **** do you think UNIX is  anyway? It's a big honkin' party of 3rd
party free tools." -- Bob Cassidy (rmcassid at uci.edu)