Re: PostgreSQL vs. MySQL

From: "Arjen van der Meijden" <acm(at)tweakers(dot)net>
To: "'Andrew Sullivan'" <andrew(at)libertyrms(dot)info>, <pgsql-performance(at)postgresql(dot)org>
Subject: Re: PostgreSQL vs. MySQL
Date: 2003-07-04 18:07:18
Message-ID: 005001c34257$1f461360$3ac15e91@acm
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-docs pgsql-general pgsql-hackers pgsql-performance

> Andrew Sullivan wrote:

> I cannot, for the life of me, understand how anyone can
> install some software which is supposed to provide meaningful
> results under production conditions, and not bother to read
> even the basic "quickstart"-type stuff that is kicking
> around.
Then please point out where it sais, in the documentation, that the
value for the shared_memory of 64 is too low and that 4000 is a nice
value to start with?

Please, also point out the part of the documentation that explains how
high the fsm-settings should be, what the impact of a lower or higher
sort_mem-setting is, what kind of value the effective_cache_size should
have and the best way to determine that.

If you can find the above in the default-documentation, like the
"getting started"-documents or the administration documentation, than be
so kind to give direct links or quotes to that. I was unable to find
that, now in a 15 minute search in the docs themselves and I have read
most part of them (in the past)...

Especially in chapter 10 "Performance hints" I was surprised not to see
such information, although it could be considered an administration
task, but there it wasn't in chapter 10 (monitoring database usage)
either.

I'm sorry to put this in a such a confronting manner, but you simply
can't expect people to search for information that they don't know the
existence of... Actually, that doesn't appear to exist, at least not on
the places you'd expect that information to be placed. I, myself, have
read Bruce's document on performance tuning, but even that document
doesn't provide the detail of information that can be read in this
mailing-list.

Having said that, this list only has 461 subscribers and I can hardly
believe that that are _all_ users of postgresql, as long as it's not the
default way of trying to gather data, it shouldn't be expected that
anyone actually tries to find his information in this list.

Anyway, I saw that there has been done some effort to create a document
that does describe such parameters, I'd be happy to see and read that :)

> There is _no secret_ that Postgres is configured as
> a toy out of the box. One presumes that DBAs are hired to do
> _some_ little bit of work.
I don't see it on the frontpage, nor in the documentation. Anyway, see
above :)

Regards,

Arjen

Btw, I've tried to tune my postgresql database using the administration
and tech documents, and saw quite a few queries run quite a lot faster
on mysql, I'll try to set up a more useful test environment and supply
this list with information to allow me to tune it to run more or less
equal to mysql. I do see 3x runs, even with the shared memory and sort
mem settings cranked up and having done little to none tuning on mysql
:)

In response to

Responses

Browse pgsql-docs by date

  From Date Subject
Next Message Arjen van der Meijden 2003-07-04 18:15:21 Re: PostgreSQL vs. MySQL
Previous Message Josh Berkus 2003-07-04 17:12:36 Re: PostgreSQL vs. MySQL

Browse pgsql-general by date

  From Date Subject
Next Message Arjen van der Meijden 2003-07-04 18:15:21 Re: PostgreSQL vs. MySQL
Previous Message john 2003-07-04 17:42:40 ANNOUNCING: Rekall V2.0.0 for Linux/UNIX, Windows, and the Sharpe family of Linux PDAs

Browse pgsql-hackers by date

  From Date Subject
Next Message Arjen van der Meijden 2003-07-04 18:15:21 Re: PostgreSQL vs. MySQL
Previous Message Tom Lane 2003-07-04 18:02:21 Re: Hitting the nfile limit

Browse pgsql-performance by date

  From Date Subject
Next Message Rod Taylor 2003-07-04 18:07:38 Re: Strange result: UNIX vs. TCP/IP sockets
Previous Message Vincent van Leeuwen 2003-07-04 17:55:12 Re: Strange result: UNIX vs. TCP/IP sockets