Message-ID: <43E2605F.5050705@rentec.com>
Date: Thu, 02 Feb 2006 14:41:19 -0500
From: Alan Stange <stange@rentec.com>
Reply-To: stange@rentec.com
Organization: Renaissance Technologies Corp.
User-Agent: Thunderbird 1.6a1 (X11/20060106)
MIME-Version: 1.0
To: "Jeffrey W. Baker" <jwbaker@acm.org>
Cc: pgsql-performance@postgresql.org
Subject: Re: Huge Data sets, simple queries
References: <C004D7AE.1B5A7%llonergan@greenplum.com>
	<1138737838.5648.4.camel@toonses.gghcwest.com>
In-Reply-To: <1138737838.5648.4.camel@toonses.gghcwest.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Jeffrey W. Baker wrote:
> On Tue, 2006-01-31 at 09:00 -0800, Luke Lonergan wrote:
>   
>> Jim,
>>
>> On 1/30/06 12:25 PM, "Jim C. Nasby" <jnasby@pervasive.com> wrote:
>>
>>     
>>> Why divide by 2? A good raid controller should be able to send read
>>> requests to both drives out of the mirrored set to fully utilize the
>>> bandwidth. Of course, that probably won't come into play unless the OS
>>> decides that it's going to read-ahead fairly large chunks of the table
>>> at a time...
>>>       
>> I've not seen one that does, nor would it work in the general case IMO.  In
>> RAID1 writes are duplicated and reads come from one of the copies.  You
>> could alternate read service requests to minimize rotational latency, but
>> you can't improve bandwidth.
>>     
>
> Then you've not seen Linux.  Linux does balanced reads on software
> mirrors.  I'm not sure why you think this can't improve bandwidth.  It
> does improve streaming bandwidth as long as the platter STR is more than
> the bus STR.
>   
FYI:   so does the Solaris Volume Manager (by default) on Solaris.   One 
can choose alternate access methods like  "First" (if the other mirrors 
are slower than the first) or "Geometric".  It's been doing this for a 
good 10 years now (back when it was called DiskSuite), so it's nothing new.

-- Alan