On Fri, Jul 16, 1999 at 10:23:33AM -0600, Sasha Pachev wrote:
> Laurent PIERRE wrote:
> > Hello,
> > I'm new on the list. I found the benchmarks results that are on the mysql
> > site, and would like to ask some questions to the people who did them.
> > If you could send me their email address, I would be very grateful.
> > Laurent.
> Remeber that the purpose of many honest benchmarks is to entertain the
> person who wrote them :-) Sometimes they are written to prove that
> product A is better than product B, those are Mindcraft-type benchmarks
> :-) Very often a slight tweak in the system may tip the benchmarked
> performance quite a bit one way or the other.
> The best way to get an idea whether a certain configuration is good for
> you or not is to build a simple system that will do operations similar
> to your target system under similar conditions and measure the
> performance yourself. The is often not an easy job, and does require
> creativity, but it is fun!
actually this is not the best way to do the evaluation it is
the only way to do it.
i (we, my partners) have seen people caught out time and time
again ... asking questions and building paper systems, a lot
like what happens here (in other places as well) from time
well anyway they get thier paper tigers assembled and do al teh
mindgame mumber schuffling, then go out and build teh cheapest
version of teh hardware model that is supposed to wrk. after it
drops dead they go looking for analysts to hide thier mess,
grin sort off.
good quality hardware is not cheap, but then again after you
have built your sample machine you can go back tothe
manufacturer and get more of teh same ... unlike the bargin
basement specials, here today gone tommorrow -- in more ones
than the obvious one, or two.
designing a database is far more than taking pen to paper and
working out the table size, file and field distribution statistics,
but then again we all know that bit don't we ???
PO Box 144, Rosebery, NSW 1445 Australia