Message-ID: <47DAD0E9.2090500@wimba.com>
Date: Fri, 14 Mar 2008 20:24:25 +0100
From: Pascal Cohen <pcohen@wimba.com>
User-Agent: Thunderbird 2.0.0.12 (X11/20080226)
MIME-Version: 1.0
To: pgsql-performance <pgsql-performance@postgresql.org>
Subject: Re: Hardware question for a DB server
References: <47D827EB.3020105@wimba.com>
	<1205349821.9048.1128.camel@archimedes>
	<Pine.GSO.4.64.0803121602410.28886@westnet.com>
In-Reply-To: <Pine.GSO.4.64.0803121602410.28886@westnet.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Greg Smith wrote:
> On Wed, 12 Mar 2008, Mark Lewis wrote:
>
>> One question that's likely going to be important depending on your
>> answers above is whether or not you're getting a battery-backed write
>> cache for that ServeRAID-8K.
>
> Apparently there's a 8k-l and an regular 8-k; the l doesn't have the 
> cache, so if this one is a regular 8-k it will have 256MB and a 
> battery. See 
> http://www.redbooks.ibm.com/abstracts/TIPS0054.html?Open#ServeRAID-8k
It is the solution with RAM and battery.
>
>> From Pascal's description of the application this system sounds like 
> overkill whether or not there's a cache.  For scaling to lots of small 
> requests, using things like using connection pooling may end up being 
> more important than worring about the disk system (the database isn't 
> big enough relative to RAM for that to be too important).
>
I agree with what you are saying. We are using Java with a pool of 
connections to access the DB. Today our database is really small 
compared to the RAM but it may evolve and even will probably grow (hope 
so which would be a good situation).

Thanks for your advices/remarks.