Message-ID: <4C238FB5.2080302@krogh.cc>
Date: Thu, 24 Jun 2010 19:02:45 +0200
From: Jesper Krogh <jesper@krogh.cc>
User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US;
	rv:1.9.1.9) Gecko/20100423 Thunderbird/3.0.4
MIME-Version: 1.0
To: Janning Vygen <vygen@kicktipp.de>
CC: pgsql-performance@postgresql.org
Subject: Re: Write performance
References: <201006241443.33296.ml@planwerk6.de>
	<alpine.DEB.2.00.1006241347180.2534@aragorn.flymine.org>
	<201006241516.05241.ml@planwerk6.de>
	<201006241545.31749.vygen@kicktipp.de>
In-Reply-To: <201006241545.31749.vygen@kicktipp.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

On 2010-06-24 15:45, Janning Vygen wrote:
> On Thursday 24 June 2010 15:16:05 Janning wrote:
>    
>> On Thursday 24 June 2010 14:53:57 Matthew Wakeling wrote:
>>      
>>> On Thu, 24 Jun 2010, Janning wrote:
>>>        
>>>> We have a 12 GB RAM machine with intel i7-975 and using
>>>> 3 disks "Seagate Barracuda 7200.11, ST31500341AS (1.5 TB)"
>>>>
>>>>          
>>> For each drive, you will be able to read/write approximately 8kB /
>>> 0.0085s, giving 941kB per second. If you have multiple processes all
>>> doing random access, then you may be able to utilise both discs and get
>>> double that.
>>>        
>> So with your calculation I have a maximum of 2MB/s random access. So i
>> really need to upgrade my disk configuration!
>>      
> i was looking at tomshardware.com and the fastest disk is
>
>     Maxtor Atlas 15K II * 8K147S0,SAS,147 GB, 16 MB Cache,15000 rpm
>
> with 5.5 ms random access time.
>
> So even if i switch to those disks i can only reach a perfomace gain of 1.5,
> right?
>
> To achieve a better disk performance by factor of ten, i need a raid-10 setup
> with 12 disks (so i have 6 raid-1 bundles). Or are there other factors with
> high end disks?
>    

Well. On the write-side, you can add in a Raid controller with Battery 
backed
write cache to not make the writes directly hit disk. This improves
the amount of writing you can do.

On the read-side you can add more memory to your server so a significant
part of your most active dataset is cached in memory.

It depends on the actual sizes and workload what gives the most benefit
for you.

-- 
Jesper