RE: Solid state disks for Oracle?

The trend, that I have seen with hardware, is that as demand goes up, more suppliers come into play and the cost comes down. Currently the cost it is to the tune of 2 millions dollars per TB. Which may not be much for larger companies. Again cost is again relative.    

  Your NFS->San gateway-> SSD is very good.    

  Mike Ault's paper is more recent that both yours and James Morle, and thereby uses more SSD. The paper is good for its experimental nature. As such, mostly oracle default parameters are being used for comparision. Tweaking the parameters and looking at the sql plans might have yielded a better SSD throughput.    

  SSD is most likely to be widespread sooner than later. I was looking at the possibility of partitioning the data, so that current data is on SSD and historical is on hard disks, and how this will affect the sql plans. As I have no place to test, most of the discussion is theoretical and I can only look at the work published by others. Hence, I sent the paper to the group, so that all may share it.    

  Joseph Amalraj

Kevin Closson <kevinc_at_polyserve.com> wrote:

From: Joseph Amalraj [mailto:joseph_at_amalrajinc.com] Sent: Thursday, March 09, 2006 1:14 PM
To: kevinc_at_polyserve.com; oracle-l_at_freelists.org Subject: RE: Solid state disks for Oracle?

Mike Ault has done a comparision which is available at http://www.superssd.com/Ault.zip

Does it really take 104 slides to point out that a solid state disk is faster than 7 SATA drives? What am I missing here?

I tried to turn this thread into one of a bit more sophistication by bringing up the fact that these things are very expensive and you can't just sit one over in the corner and get your money's worth because they are simple SAN arrays that serve up LUNS.
--

http://www.freelists.org/webpage/oracle-l

--

http://www.freelists.org/webpage/oracle-l Received on Thu Mar 09 2006 - 17:50:36 CST