From: James Mansion <james@mansionfamily.plus.com>
To: PFC <lists@peufeu.com>
Cc: Francisco Reyes <lists@stringsutils.com>
Cc: Pgsql performance <pgsql-performance@postgresql.org>
Subject: Re: RAID controllers for Postgresql on large setups
Date: Tue, 13 May 2008 21:12:58 +0100
Message-ID: <4829F64A.9090109@mansionfamily.plus.com> (raw)
In-Reply-To: <op.ua26u3iacigqcu@apollo13.peufeu.com>
References: <cone.1210644243.416947.84940.1000@zoraida.natserv.net>
<op.ua26u3iacigqcu@apollo13.peufeu.com>
PFC wrote:
> PCI limits you to 133 MB/s (theoretical), actual speed being
> around 100-110 MB/s.
Many servers do have more than one bus. You have to process that data
too so its not going to be as much of a limit as you are suggesting. It
may be possible to stream a compressed data file to the server and copy
in from that after decompression, which will free LAN bandwidth. Or
even if you RPC blocks of compressed data and decompress in the proc and
insert right there.
> On your current setup with 15K drives if you need 1 fsync per
> INSERT you won't do more than 250 per second, which is very limiting...
Well, thats 250 physical syncs. But if you have multiple insert streams
(for group commit), or can batch the rows in each insert or copy, its
not necessarily as much of a problem as you seem to be implying.
Particularly if you are doing the holding table trick.
James
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-performance@postgresql.org
Cc: james@mansionfamily.plus.com, lists@peufeu.com, lists@stringsutils.com
Subject: Re: RAID controllers for Postgresql on large setups
In-Reply-To: <4829F64A.9090109@mansionfamily.plus.com>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox