From: "Lionel" <lionel@art-informatique.com>
Subject: Re: Which hardware ?
Date: Fri, 20 Jun 2008 12:53:45 +0200
Organization: Hub.Org Networking Services
Lines: 29
Message-ID: <g3g27p$2md3$1@news.hub.org>
References: <g38epm$2pl4$1@news.hub.org>
	<20080617142501.GG68434@commandprompt.com>
	<Pine.GSO.4.64.0806171108060.7150@westnet.com>
	<dcc563d10806170842g6d039d7bsec81fb562ba9facd@mail.gmail.com>
To: pgsql-performance@postgresql.org

"Scott Marlowe" wrote:
> We had a reporting server with about 80G of data on a machine with 4G
> ram last place I worked, and it could take it a few extra seconds to
> hit the old data, but the SW RAID-10 on it made it much faster at
> reporting than it would have been with a single disk.

Would this be a nice choice ?

HP Proliant DL320 G5p  Xeon DC 3 GHz - 8 Go RAM DDR2 ECC
- 4 x 146 Go SAS 15k rpm   - RAID-10 HP Smart Array (128 Mo cache)

I finally choose to have 2 data tables:
- one with pre aggregated (dividing size by 10), unpartitionned (=the 
database they currently use)
- one with original data, yearly partitionned

I will choose before each statement which table will be used depending on 
which select/joins/where/groupby the user choosed.
The aggregated datas will allow me to maintain actual performances (and even 
improve it using the new hardware twice more powerfull).

I think lines aggregation will be handled by the java application (excel/csv 
file loaded in memory),
which will be much faster than using a trigger on insertion in the full 
table.

Thanks.