Message-ID: <493F21E9.9010009@emolecules.com>
Date: Tue, 09 Dec 2008 17:56:57 -0800
From: Craig James <craig_james@emolecules.com>
User-Agent: Thunderbird 2.0.0.18 (Macintosh/20081105)
MIME-Version: 1.0
To: justin <justin@emproshunts.com>
CC: pgsql-performance@postgresql.org
Subject: Re: Need help with 8.4 Performance Testing
References: <C5642B9E.F8A%scott@richrelevance.com>
	<22285.1228862315@sss.pgh.pa.us> <493EFD07.4080201@emproshunts.com>
In-Reply-To: <493EFD07.4080201@emproshunts.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

justin wrote:
> Tom Lane wrote:
>> Hmm ... I wonder whether this means that the current work on
>> parallelizing I/O (the posix_fadvise patch in particular) is a dead
>> end.  Because what that is basically going to do is expend more CPU
>> to improve I/O efficiency.  If you believe this thesis then that's
>> not the road we want to go down.
>>
>> 			regards, tom lane
>>
> What does the CPU/ Memory/Bus performance road map look like?
> 
> Is the IO performance for storage device for what ever it be, going to 
> be on par with the above to cause this problem?

Flash memory will become just a fourth layer in the memory caching system (on-board CPU, high-speed secondary cache, main memory, and persistent memory).  The idea of external storage will probably disappear altogether -- computers will just have memory, and won't forget anything when you turn them off.  Since most computers are 64 bits these days, all data and programs will just hang out in memory at all times, and be directly addressable by the CPU.

The distinction between disk and memory arose from the fact that disks were large, slow devices relative to "core" memory and had to be connected by long wires, hence the need for I/O subsystems.  As flash memory becomes mainstream, I expect this distinction to disappear.

Craig