Message-ID: <4CBDC72D.3080301@2ndquadrant.com>
Date: Tue, 19 Oct 2010 12:28:29 -0400
From: Greg Smith <greg@2ndquadrant.com>
User-Agent: Thunderbird 2.0.0.24 (X11/20100317)
MIME-Version: 1.0
To: Jon Nelson <jnelson+pgsql@jamponi.net>
CC: pgsql-performance@postgresql.org
Subject: Re: odd postgresql performance (excessive lseek)
References: <AANLkTikS8ptnCVJK+6u-Z3Xh8WMKdgq+T+trdxt3xgYB@mail.gmail.com>
	<AANLkTin3qV-dPeu771VLGmbiPPpeHz2oT=0t6CPtaSeS@mail.gmail.com>
	<10133.1287498969@sss.pgh.pa.us>
	<AANLkTi=RTJMHXN_zMcYdDCRnd4zDDMW=-PEB1egpCoRS@mail.gmail.com>
In-Reply-To: <AANLkTi=RTJMHXN_zMcYdDCRnd4zDDMW=-PEB1egpCoRS@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit

Jon Nelson wrote:
> That's a little harsh (it's not untrue, though).
>   

Welcome to pgsql-performance!  You can get a right answer, or a nice 
answer, but given the personalities involved it's hard to get both at 
the same time.  With this crowd, you need to be careful stating 
something you were speculating about as if you were certain of it, and 
to be certain here usually means "I read the source code".  I recommend 
writing theories as a question ("would X have sped this up?") rather 
than a statement ("X will speed this up") if you want to see gentler 
answers.

> Has any work been done on making use of shared memory for file stats
> or using fallocate (or posix_fallocate) to allocate files in larger
> chunks?
>   

Just plain old pre-allocating in larger chunks turns out to work well.  
Am hoping to get that into an official patch eventually.

-- 
Greg Smith, 2ndQuadrant US greg@2ndQuadrant.com Baltimore, MD
PostgreSQL Training, Services and Support  www.2ndQuadrant.us
Author, "PostgreSQL 9.0 High Performance"    Pre-ordering at:
https://www.packtpub.com/postgresql-9-0-high-performance/book