From: Simon Waters <simonw@zynet.net>
Organization: ZyNet Limited
To: pgsql-performance@postgresql.org
Subject: Re: Need help with 8.4 Performance Testing
Date: Tue, 9 Dec 2008 13:58:43 +0000
User-Agent: KMail/1.9.9
References: <200812071138.01215.josh@agliodbs.com>
	<Pine.GSO.4.64.0812082225470.13669@westnet.com>
	<493E6DBE.2040605@verizon.net>
In-Reply-To: <493E6DBE.2040605@verizon.net>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200812091358.43741.simonw@zynet.net>

On Tuesday 09 December 2008 13:08:14 Jean-David Beyer wrote:
> 
> and even if they can, I do not know if postgres uses that ability. I doubt
> it, since I believe (at least in Linux) a process can do that only if run
> as root, which I imagine few (if any) users do.

Disclaimer: I'm not a system programmer... 

I believe that at Linux kernel revision 2.6.8 and before processes need Posix 
capability CAP_IPC_LOCK, and 2.6.9 and after they need CAP_IPC_LOCK to lock 
more than RLIMIT_MEMLOCK.

It is a capability, so a process can run as any user assuming it is started 
with or gained the capability.

No idea if Postgres uses any of this, other than to protect security of 
certain password operations there is probably not much point. If key parts of 
your database are being paged out, get more RAM, if idle parts of your 
database are paged out, you probably could more usefully apply that RAM for 
something else.

The Varnish cache design is the place to look for enlightenment on relying on 
the kernel paging (using memory mapped files) rather than trying to do it 
yourself, but then a proxy server is a lot simpler than a RDBMS. That said, 
Varnish is fast at what it does (reverse HTTP proxy) !