Received: from localhost (unknown [200.46.204.184]) by postgresql.org (Postfix) with ESMTP id 6642B2E16F8 for ; Wed, 23 Jan 2008 21:10:12 -0400 (AST) Received: from postgresql.org ([200.46.204.71]) by localhost (mx1.hub.org [200.46.204.184]) (amavisd-maia, port 10024) with ESMTP id 20678-02-9 for ; Wed, 23 Jan 2008 21:10:00 -0400 (AST) Received: from westnet.com (westnet.com [216.187.52.2]) by postgresql.org (Postfix) with ESMTP id 13B792E3828 for ; Wed, 23 Jan 2008 20:54:27 -0400 (AST) Received: from westnet.com (localhost [127.0.0.1]) by westnet.com (8.14.0/8.14.0) with ESMTP id m0O0sOZ7020197; Wed, 23 Jan 2008 19:54:24 -0500 (EST) Received: from localhost (gsmith@localhost) by westnet.com (8.14.0/8.13.2/Submit) with ESMTP id m0O0sO9Y020194; Wed, 23 Jan 2008 19:54:24 -0500 (EST) X-Authentication-Warning: westnet.com: gsmith owned process doing -bs Date: Wed, 23 Jan 2008 19:54:24 -0500 (EST) From: Greg Smith X-X-Sender: gsmith@westnet.com To: Guy Rouillier cc: pgsql-performance@postgresql.org Subject: Re: Making the most of memory? In-Reply-To: <47979C15.3040406@burntmail.com> Message-ID: References: <4796BE84.9030803@centriccrm.com> <47979C15.3040406@burntmail.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed X-Virus-Scanned: Maia Mailguard 1.0.1 X-Archive-Number: 200801/207 X-Sequence-Number: 28446 On Wed, 23 Jan 2008, Guy Rouillier wrote: > Flash has a limited number of writes before it becomes unreliable. On > good quality consumer grade, that's about 300,000 writes, while on > industrial grade it's about 10 times that. The main advance that's made SSD practical given the write cycle limitation is increasing sophisticated wear leveling: http://en.wikipedia.org/wiki/Wear_levelling The best devices now use static wear levelling; overviews at http://en.wikipedia.org/wiki/Static_Wear_Leveling and http://www.storagesearch.com/siliconsys-art1.html The basic idea is that the number of writes to each block is tracked, and as it approaches the limit that block gets swapped with one that has been more read-only. So essentially the number of writes before failure approaches something closer to 1M x number of blocks. This means that as the size of the device goes up, so does its longevity. If you believe the hype, the combination in the increase in size of designs with these more sophisticated wear-levelling approaches has now crossed the line where it's more likely a standard moving-parts hard drive will fail first if you compare it to a similarly sized SDD doing the same job (a standard mechanical drive under heavy write load also wears out faster than one doing less work). -- * Greg Smith gsmith@gregsmith.com http://www.gregsmith.com Baltimore, MD