From: "Anibal David Acosta" <aa@devshock.com>
To: <tv@fuzzy.cz>
Cc: <pgsql-performance@postgresql.org>
References: <000601cc2761$a2e3a280$e8aae780$@devshock.com>
	<243bcc276983a6b1e2bbc783c3a5b7db.squirrel@sq.gransy.com>
In-Reply-To: <243bcc276983a6b1e2bbc783c3a5b7db.squirrel@sq.gransy.com>
Subject: Re: how much postgres can scale up?
Date: Fri, 10 Jun 2011 08:56:50 -0400
Message-ID: <000f01cc276d$df82edc0$9e88c940$@devshock.com>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Thread-Index: AQGVV9RYQGqh9nHidNYEIlkWfJ69yQFcAFsJlRlKI2A=
Content-Language: es-py

The version is Postgres 9.0
Yes, I setup the postgres.conf according to instructions in the=20
http://wiki.postgresql.org/wiki/Tuning_Your_PostgreSQL_Server


Cool, I will check this
http://wiki.postgresql.org/wiki/Logging_Difficult_Queries

Looks like great starting point to find bottleneck

But so, Is possible in excellent conditions that two connections =
duplicate the quantity of transactions per second?

Thanks!


-----Mensaje original-----
De: tv@fuzzy.cz [mailto:tv@fuzzy.cz]=20
Enviado el: viernes, 10 de junio de 2011 08:10 a.m.
Para: Anibal David Acosta
CC: pgsql-performance@postgresql.org
Asunto: Re: [PERFORM] how much postgres can scale up?

> I have a function in pgsql language, this function do some select to=20
> some tables for verify some conditions and then do one insert to a=20
> table with NO index. Update are not performed in the function
>
> When 1 client connected postgres do 180 execution per second With 2=20
> clients connected postgres do 110 execution per second With 3 clients=20
> connected postgres do 90 execution per second
>
> Finally with 6 connected clients postgres do 60 executions per second=20
> (totally 360 executions per second)
>
> While testing, I monitor disk, memory and CPU and not found any =
overload.

There's always a bottleneck - otherwise the system might run faster (and =
hit another bottleneck eventually). It might be CPU, I/O, memory, =
locking and maybe some less frequent things.

> I know that with this information you can figure out somethigns, but=20
> in normal conditions, Is normal the degradation of performance per=20
> connection when connections are incremented?
> Or should I spect 180 in the first and something similar in the second =

> connection? Maybe 170?
>
>
> The server is a dual xeon quad core with 16 GB of ram and a very fast=20
> storage The OS is a windows 2008 R2 x64

Might be, but we need more details about how the system works. On Linux =
I'd ask for output from 'iostat -x 1' and 'vmstat 1' but you're on =
Windows so there are probably other tools.

What version of PostgreSQL is this? What are the basic config values =
(shared_buffers, work_mem, effective_cache_size, ...)? Have you done =
some tuning? There's a wiki page about this:
http://wiki.postgresql.org/wiki/Tuning_Your_PostgreSQL_Server

Have you tried to log slow queries? Maybe there's one query that makes =
the whole workload slow? See this:
http://wiki.postgresql.org/wiki/Logging_Difficult_Queries

Tomas