Re: New to PostgreSQL, performance considerations

On 12/11/06, Steinar H. Gunderson <sgunderson@bigfoot.com> wrote:
> On Mon, Dec 11, 2006 at 11:17:06AM -0200, Daniel van Ham Colchete wrote:
> > I just remebered one case with MySQL. When I changed the distro from
> > Conectiva 10 (rpm-based ended brazilian distro) to Gentoo, a MySQL
> > operation that usually took 2 minutes to run, ended in 47 seconds.
>
> How do you know that this improvement had _anything_ to do with the use of
> different optimization flags? Were even the MySQL versions or configuration
> the same?
>
> > This is absolutely vage.
>
> Indeed it is.
Finally we agreed on something.

>
> > I don't have how to prove it to you.
>
> No, but you should stop making this sort of "absolutely essential" claims if
> you can't.
>
> > And I can't mesure how each factor helped: compiling glibc and Mysql with
> > good cflags, rebuilding my database in a ordered way, never kernel, etc..
>
> Exactly. So why are you attributing it to the first factor only? And why do
> you think this would carry over to PostgreSQL?
>
> Remember, anecdotal evidence isn't.

But that's exactly what I said. I'm not attributing this case to the
optimization factor. As I said there are a lot of factors involved.
The MySQL version change of a minor upgrade (from 4.1.15 to 4.1.21).

Steinar, I say I'll do the benchmark and it will be the end of the story.

>
> /* Steinar */