Re: Postgres refusing to use >1 core

From: Scott Marlowe <scott(dot)marlowe(at)gmail(dot)com>
To: Aren Cambre <aren(at)arencambre(dot)com>
Cc: Pierre C <lists(at)peufeu(dot)com>, "pgsql-performance(at)postgresql(dot)org" <pgsql-performance(at)postgresql(dot)org>
Subject: Re: Postgres refusing to use >1 core
Date: 2011-05-12 03:35:38
Message-ID: BANLkTimz1ZLu=+_H_RfoXROwXQLvTJE80Q@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

On Wed, May 11, 2011 at 9:20 PM, Aren Cambre <aren(at)arencambre(dot)com> wrote:
>> Using unix sockets, you can expect about 10-20.000 queries/s on small
>> simple selects per core, which is quite a feat. TCP adds overhead, so it's
>> slower. Over a network, add ping time.
>
> I'm talking to a Postgres on localhost, so in theory, I ought to be getting
> really good throughput, but again, the problem may be with the way C#'s
> PLINQ "multithreading in a can" is managing things.

local tcp is gonna be slower not faster than unix sockets, not faster.
But the big issue is that you need to exlpore doing the work in a
large set not iteratively. Operations on sets are often much faster
in aggregate.

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message Robert Haas 2011-05-12 03:36:19 Re: Checkpoint execution overrun impact?
Previous Message Robert Haas 2011-05-12 03:22:27 Re: DBT-5 & Postgres 9.0.3