Received: from localhost (unknown [200.46.204.183]) by postgresql.org (Postfix) with ESMTP id B1ADA64FC2E for ; Wed, 24 Sep 2008 01:25:36 -0300 (ADT) Received: from postgresql.org ([200.46.204.86]) by localhost (mx1.hub.org [200.46.204.183]) (amavisd-maia, port 10024) with ESMTP id 09082-03 for ; Wed, 24 Sep 2008 01:25:31 -0300 (ADT) X-Greylist: from auto-whitelisted by SQLgrey-1.7.6 Received: from wf-out-1314.google.com (wf-out-1314.google.com [209.85.200.169]) by postgresql.org (Postfix) with ESMTP id CE94164FC24 for ; Wed, 24 Sep 2008 01:25:28 -0300 (ADT) Received: by wf-out-1314.google.com with SMTP id 25so2639043wfc.28 for ; Tue, 23 Sep 2008 21:25:27 -0700 (PDT) Received: by 10.142.50.15 with SMTP id x15mr2328151wfx.342.1222230326931; Tue, 23 Sep 2008 21:25:26 -0700 (PDT) Received: by 10.143.39.19 with HTTP; Tue, 23 Sep 2008 21:25:26 -0700 (PDT) Message-ID: Date: Tue, 23 Sep 2008 21:25:26 -0700 From: "Scott Carey" To: "Bruce Momjian" Subject: Re: Intel's X25-M SSD Cc: "Greg Smith" , "Merlin Moncure" , pgsql-performance@postgresql.org In-Reply-To: <200809240324.m8O3Ovn06515@momjian.us> MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_Part_137781_12768976.1222230326890" References: <200809240324.m8O3Ovn06515@momjian.us> X-Google-Sender-Auth: 6374fee52b6b97e1 X-Virus-Scanned: Maia Mailguard 1.0.1 X-Spam-Status: No, hits=0.001 tagged_above=0 required=5 tests=HTML_MESSAGE=0.001 X-Spam-Level: X-Archive-Number: 200809/212 X-Sequence-Number: 31525 ------=_Part_137781_12768976.1222230326890 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline A fantastic review on this issue appeared in July: http://www.alternativerecursion.info/?p=106 And then the same tests on a RiData SSD show that they are the same drive with the same characteristics: http://www.alternativerecursion.info/?p=276 Most blamed it on MLC being "slow" to write compared to SLC. Technically, it is slower, but not by a whole lot -- we're talking a low level difference of tens of microseconds. A 250ms latency indicates an issue with the controller chip. SLC and MLC share similar overall performance characteristics at the millisecond level. The truth is that MLC designs were low cost designs without a lot of investment in the controller chip. The SLC designs were higher cost designs that focused early on on making smarter and more expensive controllers. SLC will always have an advantage, but it isn't going to be by several orders of magnitude like it was before Intel's drive appeared. Its going to be by factors of ~2 to 4 on random writes in the long run. However, for all flash based SSD devices, there are design tradeoffs to make. Maximizing writes sacrifices reads, maximizing random access performance reduces streaming performance and capacity. We'll have a variety of devices with varying characteristics optimal for different tasks. On Tue, Sep 23, 2008 at 8:24 PM, Bruce Momjian wrote: > Greg Smith wrote: > > On Mon, 8 Sep 2008, Merlin Moncure wrote: > > > > > What's interesting about the X25 is that they managed to pull the > > > numbers they got out of a MLC flash product. They managed this with a > > > DRAM buffer and the custom controller. > > > > I finally found a good analysis of what's wrong with most of the cheap > MLC > > drives: > > > > http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3403&p=7 > > > > 240ms random write latency...wow, no wonder I keep hearing so many > reports > > of cheap SSD just performing miserably. JMicron is one of those > companies > > I really avoid, never seen a design from them that wasn't cheap junk. > > Shame their awful part is in so many of the MLC flash products. > > I am surprised it too so long to identify the problem. > > -- > Bruce Momjian http://momjian.us > EnterpriseDB http://enterprisedb.com > > + If your life is a hard drive, Christ can be your backup. + > > -- > Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org) > To make changes to your subscription: > http://www.postgresql.org/mailpref/pgsql-performance > ------=_Part_137781_12768976.1222230326890 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline
A fantastic review on this issue appeared in July:
http://www.alternativ= erecursion.info/?p=3D106
And then the same tests on a RiData SSD sho= w that they are the same drive with the same characteristics:
http://www.altern= ativerecursion.info/?p=3D276

Most blamed it on MLC being "s= low" to write compared to SLC.  Technically, it is slower, but no= t by a whole lot -- we're talking a low level difference of tens of mic= roseconds.  A 250ms latency indicates an issue with the controller chi= p.  SLC and MLC share similar overall performance characteristics at t= he millisecond level.  The truth is that MLC designs were low cost des= igns without a lot of investment in the controller chip.  The SLC desi= gns were higher cost designs that focused early on on making smarter and mo= re expensive controllers.  SLC will always have an advantage, but it i= sn't going to be by several orders of magnitude like it was before Inte= l's drive appeared.  Its going to be by factors of ~2 to 4 on rand= om writes in the long run.  However, for all flash based SSD devices, = there are design tradeoffs to make.  Maximizing writes sacrifices read= s, maximizing random access performance reduces streaming performance and c= apacity.  We'll have a variety of devices with varying characteris= tics optimal for different tasks.

On Tue, Sep 23, 2008 at 8:24 PM, Bruce Momji= an <bruce@momjian.= us> wrote:
Greg Smith wrote:
> On Mon, 8 Sep 2008, Merlin Moncure wrote:
>
> > What's interesting about the X25 is that they managed to pull= the
> > numbers they got out of a MLC flash product.  They managed t= his with a
> > DRAM buffer and the custom controller.
>
> I finally found a good analysis of what's wrong with most of the c= heap MLC
> drives:
>
> http://www.anandtech.com/cpuchipsets/i= ntel/showdoc.aspx?i=3D3403&p=3D7
>
> 240ms random write latency...wow, no wonder I keep hearing so many rep= orts
> of cheap SSD just performing miserably.  JMicron is one of those = companies
> I really avoid, never seen a design from them that wasn't cheap ju= nk.
> Shame their awful part is in so many of the MLC flash products.

I am surprised it too so long to identify the problem.

--
 Bruce Momjian  <bruce@mo= mjian.us>        http://momjian.us
 EnterpriseDB                =             http://enterprisedb.com

 + If your life is a hard drive, Christ can be your backup. +

--
Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance

------=_Part_137781_12768976.1222230326890--